feat(software): add professional software job center

AGENTS.md

@@ -36,8 +36,11 @@
 
 ## 文档与版本
 
-- 版本号、`CHANGELOG.md`、`PROGRESS.md` 在 push 前统一更新一次，不按每个 commit 更新。
-- push 前更新 `PROGRESS.md`：补“已完成关键能力”条目，状态表变化随之更新，新增或删除模块时同步文件清单。
+- 开发 push 与正式发布分离：功能开发期间允许持续 commit / push，不因此提升版本号，也不新增已发布的 `CHANGELOG.md` 数字版本条目。
+- 尚未发布但需要预先整理的用户文案写在 `CHANGELOG.md` 顶部 `## Unreleased`；该区不会被前端更新日志接口解析。正式发布时再把它改成 `## <版本> — <日期>`。
+- 版本号与用户版 `CHANGELOG.md` 只在功能稳定、准备上线时通过单独的 release commit 统一更新；同一次发布中校准 `PROGRESS.md`，不按每个开发 commit 更新。
+- 阶段性成果可随开发更新 `PROGRESS.md`；正式发布前补“已完成关键能力”条目，状态表变化随之更新，新增或删除模块时同步文件清单。
+- 未完成且不能让线上用户接触的功能，应在独立功能分支开发、待稳定后合并生产分支；若必须提前合并或部署，则使用默认关闭的 feature flag，并限制为管理员或测试账号启用。
 - 版本号唯一事实源是 `core/__init__.py::__version__`：
   - patch：bug 修复、重构、调参、新 skill、样式；
   - minor：成批新功能或明显对外行为变化；
@@ -59,4 +62,3 @@
 ## 领域语境
 
 主要使用方是中国建筑材料科学研究总院，核心语境是无机非金属材料研发与生产，包括水泥、混凝土、玻璃、陶瓷、耐火材料和新型建材。典型任务是配方研发、性能测试、XRD/SEM/热分析、实验数据建模，以及申报书、调研报告、专利和论文写作。默认按材料研发而非建筑施工、结构计算或 BIM 理解需求。
-

CHANGELOG.md

@@ -3,7 +3,13 @@
 > 面向用户的更新说明,由 `GET /v1/changelog` 解析后在前端「点版本号」弹层展示。
 > 维护口径:只记**用户可感知**的变化("能干什么了"而不是"改了哪个模块"),纯内部重构 / 修复不记,
 > 所以不是每个版本号都有条目。条目格式 `## <版本> — <日期>`,新条目加在最上面。
+> 开发中的用户文案可先写入 `## Unreleased`；该区不会被前端解析，正式发布时再替换为数字版本和日期。
 > 工程口径的完整记录见 `PROGRESS.md` / git log。
+
+## Unreleased
+
+- 新增专业软件 Job 中心：Agent 可提交、查询和停止 Windows Node 上的受控软件任务；用户可在右下角跨对话查看进度、收到完成通知，并回到原对话分析结果。
+
 ## 0.65.2 — 2026-08-13
 
 - 粘贴、拖拽或选择附件发送后，对话中的用户消息会保留附件卡片；图片同时显示可点击放大的缩略图，刷新历史对话后也不会消失。

DESIGN.md

@@ -460,11 +460,21 @@ scheduled_jobs(§8.5)   channel_bindings(§8.7,判别列+JSONB)
 
 第一阶段以 `docs/windows-node-mvp-intranet.md` 为实现契约：Windows Node 只作为受控执行节点，通过出站 HTTP/WS 主动连接 zcbot；首批能力固定为 `origin.plot@v1`。长期方案中的 mTLS、Service/DesktopRunner 双进程、完整租约与多节点调度暂不进入 MVP，但 URL path、Node ID、Bearer Header 和任务协议保留原位升级空间。
 
-云端控制面使用独立的 `compute_node_enrollments` 与 `compute_nodes`，不复用用户外部系统连接。管理员创建的一次性注册码具有 128 bit 随机熵，数据库只保存 SHA-256 摘要；节点注册在行锁事务中校验有效期、预期名称和允许能力，成功后原子消费。每个节点获得独立高熵 Token，数据库只保存 bcrypt 强哈希，明文仅在注册响应出现一次。
+云端控制面使用独立的 `software_node_enrollments` 与 `software_nodes`，不复用用户外部系统连接。管理员创建的一次性注册码具有 128 bit 随机熵，数据库只保存 SHA-256 摘要；节点注册在行锁事务中校验有效期、预期名称和允许能力，成功后原子消费。每个节点获得独立高熵 Token，数据库只保存 bcrypt 强哈希，明文仅在注册响应出现一次。
 
-Node 通过 `Authorization: Bearer` 与 `X-Node-Id` 建立 `/v1/compute/nodes/connect` WebSocket。进程内 Connection Manager 保证同一节点单活，新连接关闭旧连接；`hello`/`heartbeat` 更新版本、容量、软件健康与最后在线时间。管理员禁用节点时先持久化禁用态，再关闭现有连接；断线收尾不得覆盖禁用态。当前单活只覆盖单 Web 进程，生产启用多实例前必须增加 Redis/PG fencing 或将 Node API 固定路由到单一控制面实例。
+Node 通过 `Authorization: Bearer` 与 `X-Node-Id` 建立 `/v1/software-nodes/connect` WebSocket。进程内 Connection Manager 保证同一节点单活，新连接关闭旧连接；`hello`/`heartbeat` 更新版本、容量、软件健康与最后在线时间。管理员禁用节点时先持久化禁用态，再关闭现有连接；断线收尾不得覆盖禁用态。当前单活只覆盖单 Web 进程，生产启用多实例前必须增加 Redis/PG fencing 或将 Node API 固定路由到单一控制面实例。
 
-首批只落注册、认证、心跳、状态与禁用基础链路。`compute_jobs`、任务 offer/accept、Origin Worker、输入输出传输、重连对账和 Token 轮换属于后续垂直闭环，不以任意命令或脚本接口临时代替。
+第二阶段已增加 `software_jobs`（专业软件任务）账本与 `origin.plot@v1` 的 offer/accept 骨架。用户只能在本人 task 下以幂等键提交固定 schema；云端规范化请求并记录 SHA-256，按当前进程真实在线、能力匹配、健康且有空闲 slot 的 Node 创建短期 offer。Node 再次校验 schema、图形类型和输出格式，使用 write-through、flush 与原子 rename 先落本机任务目录，再回 `job_accept`；重复 job 只有 digest 一致才接受。过期或发送失败的 offer 回到队列，lease、Node 和 digest 不匹配的响应被拒绝。Node 接收后云端进入 `dispatched` 而非 `running`，并将 slot 降为 0；只有固定 Worker 真正启动后才进入 `origin_running`。
+
+第三阶段补齐输入下载与恢复状态协议：`input_id` 固定为用户已有 artifact UUID，提交时快照文件名、大小和 SHA-256，只允许 CSV/XLSX/JSON 且不超过 100 MiB。Node 以自身 Bearer 身份访问任务绑定的只读下载端点，流式写入本 job 的 `input/`，同时限制声明大小并校验 SHA-256，完成后原子 rename；不暴露工作区路径。Node 会原子读取/补报 `terminal.json`，断线后云端把活动任务标记 `disconnected` 并保留 Node/lease，重连按 job、lease、digest 恢复下载或幂等补报终态，不自动重派。
+
+第四阶段落地固定 Origin Worker：Node 仅从管理员安装的固定 Python 运行时启动随程序发布的 `worker.py`，参数只有本机 job 目录；请求不能指定脚本、解释器或文件路径。Worker 使用 `originpro` 生成 OPJU、PNG、SVG、PDF、plot spec 和 provenance，校验产物签名并原子写入终态；当前受控图形仅含 line、scatter、line_scatter 和双栏出版布局。进程内 pipeline 按 job 去重，并脱离单次 WebSocket 的取消令牌运行；连接中断只延迟状态/终态上报。Node 进程若在 Worker 启动后重启，则保守失败而不重复驱动 Origin，避免无法证明的双执行。
+
+第五阶段完成输出上传与发布：Node 只按固定 manifest ID 逐项流式 PUT，并携带 Node、lease、request digest 与内容摘要；云端重新绑定任务身份，不信 Node 提供的路径或媒体类型。文件先进入用户根下隐藏暂存区，固定文件名、单文件/总大小和 SHA-256 全部验证后，目录级原子移动到 `<working_dir>/origin/<job_id>/`，再登记平台 artifact UUID 并写成功终态。重复 PUT、complete 和重连均按摘要幂等；部分上传不可见，只有完整集合才能发布。
+
+第六阶段增加用户级 Job 中心与 Agent typed tools。`software_capability_list` 只暴露固定能力及当前在线空闲节点数，`software_job_submit/status/cancel` 在构造时绑定当前 user/task，模型不能跨用户或跨对话指定归属。右下角 Job 中心按用户聚合各对话任务，活动期短轮询、空闲期降频；终态变化通知用户，成功任务可回到原对话发起分析。取消采用协作协议：未派发任务直接终止，已派发任务先进入 `cancelling`，云端通过 WebSocket 发送并在心跳时重放 `job_cancel`，Node 杀死固定 Worker 进程树后回报 `cancelled`；终态写入仍由云端账本裁决。
+
+后续仍需实现 Token 轮换；不得以任意命令或脚本接口临时代替。当前 Job 中心采用轮询而非用户事件推送，单活与 offer 选择仍只覆盖单 Web 进程；生产启用多实例前必须增加 Redis/PG fencing 或固定路由到单一控制面实例。
 
 ---
 

PROGRESS.md

@@ -2,7 +2,7 @@
 
 > 配合 `DESIGN.md`。本文件只记 phase 状态、决策偏差、文件量、下一步。每条 1-2 句:做了啥 + 关键判断;细节查 `git log` / `git diff` / `DESIGN §7.9`。
 
-最后更新:2026-08-13(用户消息结构化附件与图片预览,bump 0.65.2)
+最后更新:2026-08-13(专业软件 Job 中心开发中,未发版)
 
 ---
 
@@ -22,6 +22,8 @@
 
 ### 2026-08-13
 
+- **08-13 / Unreleased / 专业软件 Job 中心**:Agent 新增固定能力发现、提交、状态和取消工具，后端提供用户级跨对话任务列表与协作取消，Windows Node 可终止固定 Worker 进程树并幂等回报取消终态；Web 右下角展示活动/最近任务、进度、完成通知、停止与回到原对话分析入口。相关 Python 78 项、JavaScript 语法、Python 编译、.NET build 与 diff 检查通过；数据库保持在 0031，未连库、未执行 migration。
+
 - **08-13 / 0.65.2 / 用户消息结构化附件 + 对话内图片预览**:新增 0031 `messages.attachment_refs`，新客户端将附件作为结构化字段发送，数据库正文仅保留用户自然语言；后端按 task working_dir 校验路径并在内存模型上下文中补附件提示，旧客户端与历史正文标记继续兼容。用户消息即时态和历史态统一展示附件 chip，图片额外显示可点击缩略图；相关 Python 34 项、全部前端 Node 26 项、Python/JavaScript 语法、Alembic 单 head 与 diff 检查通过，未连接或迁移生产 DB。
 
 ### 2026-08-12
@@ -341,7 +343,7 @@ core/llm_transport.py      438   ← wire 层健壮性:畸形/吐空检测+留
 core/tool_registry.py      264   ← 声明式工具注册表((组名,gate,factory);secret/host 工具按实际能力 gate)
 core/context.py             95   ← LLM 调用前压缩旧 tool / load_skill 消息(带压力门槛),保 tool_call 协议字段
 core/external_systems/*.py       ← 外部系统目录/用户授权/凭据加密 + 通用 OpenAPI/MCP connector
-core/compute_nodes.py            ← Windows Node 注册码、身份认证与运行状态
+core/software_nodes.py            ← Windows Node 注册码、身份认证与运行状态
 core/sinks.py              101
 core/paths.py               50   ← task_dir db form 归一
 core/probe.py              243
@@ -363,7 +365,7 @@ tools/{base,output,fs,shell,run_python,skill_tool,skill_authoring,media_common,s
 main.py                    ~210  ← 入口:web / db / probe / user / sandbox check
 db/migrations/versions/    0001-0030
 web/app.py                 ~210  ← 工厂 + lifespan 编排(07-23 拆分;路由在 routers/,协程在 background 等)
-web/routers/*.py                 ← 含 external_systems 用户连接与 compute_nodes 节点路由
+web/routers/*.py                 ← 含 external_systems 用户连接与 software_nodes 节点路由
 web/{background,scheduler_runner,wechat_runner}.py ← lifespan 后台协程按域析出
 web/{runs,common,schemas,model_gate,userfiles}.py  ← BG worker/共享 helper/请求体/档位门控/路径安全
 web/auth.py                ~190  ← 邮箱密码 + platform_key → JWT

RUN.md

@@ -538,6 +538,18 @@ sudo bash deploy/update_bluegreen.sh
 > 蓝绿部署机(生产,已 bootstrap)一律用 `sudo bash deploy/update_bluegreen.sh`(见上节)。
 > 下面的 `update.sh` 是单实例版:host backend 机 / 内网机 / 未切蓝绿的部署继续用。
 
+开发 push 与正式发布是两个动作：开发期间可以持续 push，但不提升
+`core/__init__.py::__version__`，也不新增 `CHANGELOG.md` 的数字版本条目。需要提前整理的
+用户文案写在 `## Unreleased`，前端不会展示。功能稳定后创建单独的 release commit，统一完成：
+
+1. 将 `## Unreleased` 改为 `## <版本> — <日期>`，只保留用户可感知的变化；
+2. 更新 `core/__init__.py::__version__`；
+3. 校准 `PROGRESS.md`，并按变更类型同步本文件或 `DESIGN.md`；
+4. 部署并验证 `/healthz` 与 `/v1/changelog`。
+
+未完成且不能向线上用户开放的功能应留在功能分支，稳定后再合并生产分支；必须提前部署时，
+使用默认关闭且仅向管理员或测试账号开放的 feature flag。
+
 **一把梭(单实例):`deploy/update.sh`**
 
 ```bash
@@ -1042,14 +1054,16 @@ sudo xfs_quota -x -c "limit -p bhard=10g zcbot_<user_uuid>" /opt
 
 ### Windows Node 内网 MVP（开发中）
 
-先执行 `alembic upgrade head` 创建 `compute_node_enrollments` 和 `compute_nodes`。不要在未确认目标数据库时运行迁移；本机 `.env` 的 `ZCBOT_DB_URL` 可能是生产隧道。
+先执行 `alembic upgrade head` 创建 `software_node_enrollments`、`software_nodes` 和 `software_jobs`。不要在未确认目标数据库时运行迁移；本机 `.env` 的 `ZCBOT_DB_URL` 可能是生产隧道。
 
 云端当前提供：
 
-- 管理员 `POST /v1/admin/compute-node-enrollments` 创建一次性注册码；
-- Node `POST /v1/compute/nodes/enroll` 注册并一次性取得 `node_id`、`node_token`；
-- Node 携带 `Authorization: Bearer <node_token>` 和 `X-Node-Id` 连接 `WS /v1/compute/nodes/connect`；
-- 管理员 `GET /v1/admin/compute-nodes` 查看节点，`PATCH /v1/admin/compute-nodes/{node_id}` 启停节点，`DELETE /v1/admin/compute-nodes/{node_id}` 永久删除节点身份。
+- 管理员 `POST /v1/admin/software-node-enrollments` 创建一次性注册码；
+- Node `POST /v1/software-nodes/enroll` 注册并一次性取得 `node_id`、`node_token`；
+- Node 携带 `Authorization: Bearer <node_token>` 和 `X-Node-Id` 连接 `WS /v1/software-nodes/connect`；
+- 用户或 Agent 通过 `POST /v1/tasks/{task_id}/software-jobs` 提交专业软件任务；
+- 用户通过 `GET /v1/software-jobs` 查看本人跨对话任务，可用 `task_id`、`active_only` 和 `limit` 筛选，`POST /v1/software-jobs/{job_id}/cancel` 请求停止；
+- 管理员 `GET /v1/admin/software-nodes` 查看节点，`PATCH /v1/admin/software-nodes/{node_id}` 启停节点，`DELETE /v1/admin/software-nodes/{node_id}` 永久删除节点身份。
 
 Node API 只能绑定受控内网地址并由安全组限制来源 IP。当前 HTTP/WS 链路不加密；跨安全域、公网或不可信终端接入前，必须先升级 HTTPS/WSS。多 Web 实例部署时，Node API 暂时固定路由到单一实例，直至 Connection Manager 增加跨实例 fencing。
 
@@ -1064,11 +1078,21 @@ windows-node/Zcbot.WindowsNode/bin/Debug/net10.0-windows/Zcbot.WindowsNode.exe e
 windows-node/Zcbot.WindowsNode/bin/Debug/net10.0-windows/Zcbot.WindowsNode.exe
 ```
 
+Origin Worker 使用独立固定 Python 运行时。先确认该交互式 Windows 账号已安装并可启动 Origin/OriginPro，再由管理员安装运行时（不要复用服务端 `.venv`）：
+
+```powershell
+windows-node\install-origin-runtime.ps1 -BootstrapPython D:\programs\Python312\python.exe
+```
+
+默认解释器为 `%ProgramData%\Zcbot\WindowsNode\runtimes\origin\python.exe`。如需放在其他受管目录，设置机器级 `ZCBOT_ORIGIN_PYTHON` 为绝对 `python.exe` 路径后重启 Node。运行时固定依赖见 `windows-node/origin-worker/requirements.txt`；任务请求无权选择解释器、脚本或路径。当前 Worker 支持 CSV/XLSX/JSON 输入，`line`、`scatter`、`line_scatter` 与 OPJU/PNG/SVG/PDF 输出。成功产物由 Node 流式上传，全部校验通过后发布到任务工作目录 `origin/<job_id>/`；上传中断会在重连时幂等续传。
+
+Web 用户登录后，右下角 Job 中心会聚合本人最近任务。活动任务约 4 秒刷新一次，空闲时降为约 30 秒；停止已派发任务是协作取消，状态先显示“正在停止”，Node 在线时立即接收，断线后在下次连接或心跳时重放。Agent 可调用 `software_capability_list`、`software_job_submit`、`software_job_status` 和 `software_job_cancel`；提交工具只创建固定 schema 的持久任务，不会阻塞当前对话等待完成。
+
 注册配置写入 `%ProgramData%\Zcbot\WindowsNode\node.json`；Token 使用 DPAPI `LocalMachine` 加密，ACL 仅允许注册账号和 `SYSTEM`。应始终用同一专用 Windows 账号注册并运行 Node。当前 MVP 以该账号的登录后计划任务启动，不安装 Windows Service。
 
 直接双击 EXE 启动托盘 UI：红点为未注册/身份失效，黄点为连接中，绿点为在线；双击托盘图标打开配置窗。原 CLI 注册入口继续保留，无 UI 模式使用 `Zcbot.WindowsNode.exe run --headless`。
 
-经 nginx 反代时，`/v1/compute/nodes/connect` 必须单独透传 WebSocket Upgrade/Connection 头并设置长连接超时，配置见 `deploy/nginx/zcbot.conf.example`。若注册成功后节点持续显示“连接中断，等待重连”，先用 WebSocket 握手检查该路径；返回普通 HTTP 404 通常表示请求落入了清空 `Connection` 头的默认 location。
+经 nginx 反代时，`/v1/software-nodes/connect` 必须单独透传 WebSocket Upgrade/Connection 头并设置长连接超时，配置见 `deploy/nginx/zcbot.conf.example`。若注册成功后节点持续显示“连接中断，等待重连”，先用 WebSocket 握手检查该路径；返回普通 HTTP 404 通常表示请求落入了清空 `Connection` 头的默认 location。
 
 
 - **入口**:`main.py`(`web / db / probe / user`)→ `core/agent_builder.py::build_agent`

core/artifact_lifecycle.py

@@ -59,7 +59,7 @@ def register_published_artifacts(
                 continue
             current_path = _rel(root, path)
             label = str(ref.get("label") or "")
-            media_type = mimetypes.guess_type(path.name)[0]
+            media_type = str(ref.get("media_type") or "") or mimetypes.guess_type(path.name)[0]
             size_bytes = path.stat().st_size
             content_sha256 = _hash_file(path)
             statement = pg_insert(Artifact).values(

core/software_jobs.py

@@ -0,0 +1,757 @@
+"""专业软件任务的校验、幂等持久化和 offer 状态机。"""
+
+from __future__ import annotations
+
+import json
+import re
+from datetime import datetime, timedelta, timezone
+from hashlib import sha256
+from uuid import UUID, uuid4
+
+from sqlalchemy import desc, select
+from sqlalchemy.exc import IntegrityError
+
+from core.software_nodes import SUPPORTED_CAPABILITIES
+from core.storage.engine import session_scope
+from core.storage.models import Artifact, SoftwareJob, SoftwareNode, Task
+
+OFFER_SECONDS = 60
+ALLOWED_PLOT_TYPES = frozenset(
+    {"line", "scatter", "line_scatter"}
+)
+ALLOWED_OUTPUT_FORMATS = frozenset({"opju", "png", "svg", "pdf"})
+ALLOWED_INPUT_SUFFIXES = frozenset({".csv", ".xlsx", ".json"})
+MAX_INPUT_BYTES = 100 * 1024 * 1024
+MAX_OUTPUT_ARTIFACT_BYTES = 256 * 1024 * 1024
+class SoftwareJobError(Exception):
+    pass
+
+
+MAX_OUTPUT_TOTAL_BYTES = 512 * 1024 * 1024
+OUTPUT_ARTIFACTS = {
+    "project": ("project.opju", "application/x-origin-project", "opju"),
+    "figure_png": ("figure.png", "image/png", "png"),
+    "figure_svg": ("figure.svg", "image/svg+xml", "svg"),
+    "figure_pdf": ("figure.pdf", "application/pdf", "pdf"),
+    "plot_spec": ("plot-spec.json", "application/json", None),
+    "provenance": ("provenance.json", "application/json", None),
+}
+
+
+def _has_only(value: dict, fields: set[str]) -> bool:
+    return set(value).issubset(fields)
+
+
+def _canonical_request(request: dict) -> tuple[dict, str]:
+    if not isinstance(request, dict) or set(request) != {"schema_version", "input", "plot", "output"}:
+        raise SoftwareJobError("invalid origin plot request fields")
+    if request.get("schema_version") != 1:
+        raise SoftwareJobError("unsupported origin plot schema version")
+    input_spec = request.get("input")
+    plot = request.get("plot")
+    output = request.get("output")
+    if not all(isinstance(item, dict) for item in (input_spec, plot, output)):
+        raise SoftwareJobError("origin plot request sections must be objects")
+    if not _has_only(input_spec, {"input_id", "sheet"}):
+        raise SoftwareJobError("unsupported origin input fields")
+    try:
+        UUID(str(input_spec.get("input_id") or ""))
+    except ValueError as exc:
+        raise SoftwareJobError("input.input_id must be an artifact UUID") from exc
+    if "sheet" in input_spec and (
+        not isinstance(input_spec["sheet"], str) or not 1 <= len(input_spec["sheet"]) <= 128
+    ):
+        raise SoftwareJobError("input.sheet must be a string")
+    if not _has_only(
+        plot,
+        {"type", "x", "y", "template", "title", "x_axis", "y_axis", "legend", "error_bars"},
+    ):
+        raise SoftwareJobError("unsupported origin plot fields")
+    if plot.get("type") not in ALLOWED_PLOT_TYPES:
+        raise SoftwareJobError("unsupported origin plot type")
+    if "title" in plot and (
+        not isinstance(plot["title"], str) or len(plot["title"]) > 500
+    ):
+        raise SoftwareJobError("plot.title must be a string")
+    if plot.get("template", "publication_double_column") != "publication_double_column":
+        raise SoftwareJobError("unsupported origin plot template")
+    x_column = plot.get("x")
+    y_columns = plot.get("y")
+    if not isinstance(x_column, str) or not 1 <= len(x_column) <= 128:
+        raise SoftwareJobError("plot.x must be a column name")
+    if isinstance(y_columns, str):
+        y_columns = [y_columns]
+    if (
+        not isinstance(y_columns, list)
+        or not 1 <= len(y_columns) <= 16
+        or len(y_columns) != len(set(y_columns))
+        or any(not isinstance(item, str) or not 1 <= len(item) <= 128 for item in y_columns)
+    ):
+        raise SoftwareJobError("plot.y must contain 1 to 16 unique column names")
+    for axis_name in ("x_axis", "y_axis"):
+        axis = plot.get(axis_name)
+        if axis is not None and (
+            not isinstance(axis, dict)
+            or not _has_only(axis, {"title", "unit", "scale"})
+            or axis.get("scale", "linear") != "linear"
+            or any(
+                name in axis and not isinstance(axis[name], str)
+                for name in ("title", "unit")
+            )
+        ):
+            raise SoftwareJobError(f"invalid {axis_name}")
+    legend = plot.get("legend")
+    if legend is not None and (
+        not isinstance(legend, dict)
+        or not _has_only(legend, {"enabled", "position"})
+        or ("enabled" in legend and not isinstance(legend["enabled"], bool))
+        or legend.get("enabled", True) is not True
+        or legend.get("position", "top_right") != "top_right"
+    ):
+        raise SoftwareJobError("invalid plot.legend")
+    if plot.get("error_bars") is not None:
+        raise SoftwareJobError("error bars are not supported in origin.plot@v1")
+    if not _has_only(output, {"formats", "dpi", "capture_screenshots", "record_video"}):
+        raise SoftwareJobError("unsupported origin output fields")
+    if any(
+        name in output and not isinstance(output[name], bool)
+        for name in ("capture_screenshots", "record_video")
+    ):
+        raise SoftwareJobError("origin output capture flags must be boolean")
+    if output.get("record_video", False):
+        raise SoftwareJobError("origin video recording is not supported")
+    dpi = output.get("dpi", 300)
+    if not isinstance(dpi, int) or isinstance(dpi, bool) or not 72 <= dpi <= 1200:
+        raise SoftwareJobError("output.dpi must be between 72 and 1200")
+    formats = output.get("formats")
+    if (
+        not isinstance(formats, list)
+        or not formats
+        or len(formats) != len(set(formats))
+        or any(item not in ALLOWED_OUTPUT_FORMATS for item in formats)
+    ):
+        raise SoftwareJobError("output.formats contains unsupported values")
+    encoded = json.dumps(request, ensure_ascii=False, sort_keys=True, separators=(",", ":"))
+    if len(encoded.encode("utf-8")) > 256 * 1024:
+        raise SoftwareJobError("origin plot request is too large")
+    normalized = json.loads(encoded)
+    return normalized, sha256(encoded.encode("utf-8")).hexdigest()
+
+
+def _job_dict(row: SoftwareJob) -> dict:
+    return {
+        "job_id": str(row.job_id),
+        "task_id": str(row.task_id),
+        "capability": row.capability,
+        "request_digest": row.request_digest,
+        "node_id": str(row.node_id) if row.node_id else None,
+        "status": row.status,
+        "stage": row.stage,
+        "progress": row.progress,
+        "metrics": row.metrics,
+        "error": row.error,
+        "artifact_manifest": row.artifact_manifest,
+        "created_at": row.created_at.isoformat() if row.created_at else None,
+        "started_at": row.started_at.isoformat() if row.started_at else None,
+        "terminal_at": row.terminal_at.isoformat() if row.terminal_at else None,
+    }
+
+
+def list_jobs(
+    user_id: UUID,
+    *,
+    task_id: UUID | None = None,
+    active_only: bool = False,
+    limit: int = 50,
+) -> list[dict]:
+    """列出用户的软件任务；用于全局 Job 中心和 Agent 查询。"""
+    limit = max(1, min(int(limit), 100))
+    with session_scope() as session:
+        statement = (
+            select(SoftwareJob, Task.name, SoftwareNode.name)
+            .join(Task, Task.task_id == SoftwareJob.task_id)
+            .outerjoin(SoftwareNode, SoftwareNode.node_id == SoftwareJob.node_id)
+            .where(SoftwareJob.user_id == user_id)
+        )
+        if task_id is not None:
+            statement = statement.where(SoftwareJob.task_id == task_id)
+        if active_only:
+            statement = statement.where(
+                SoftwareJob.status.in_(
+                    {
+                        "queued",
+                        "offered",
+                        "dispatched",
+                        "running",
+                        "disconnected",
+                        "cancelling",
+                    }
+                )
+            )
+        rows = session.execute(
+            statement.order_by(
+                desc(SoftwareJob.created_at), desc(SoftwareJob.job_id)
+            ).limit(limit)
+        ).all()
+        results: list[dict] = []
+        for job, task_name, node_name in rows:
+            item = _job_dict(job)
+            item.update(
+                {
+                    "task_name": task_name,
+                    "node_name": node_name,
+                    "input": job.input_manifest,
+                    "request_summary": _request_summary(job),
+                }
+            )
+            results.append(item)
+        return results
+
+
+def _request_summary(job: SoftwareJob) -> dict:
+    plot = job.request.get("plot") or {}
+    output = job.request.get("output") or {}
+    return {
+        "display_name": (
+            "Origin 科研绘图"
+            if job.capability == "origin.plot@v1"
+            else job.capability
+        ),
+        "title": str(plot.get("title") or ""),
+        "formats": list(output.get("formats") or []),
+    }
+
+
+def request_job_cancel(user_id: UUID, job_id: UUID) -> tuple[dict, dict | None]:
+    """持久化取消意图；queued 直接终止，已分派任务返回 Node 消息。"""
+    now = datetime.now(timezone.utc)
+    with session_scope() as session:
+        job = session.execute(
+            select(SoftwareJob).where(
+                SoftwareJob.job_id == job_id,
+                SoftwareJob.user_id == user_id,
+            ).with_for_update()
+        ).scalar_one_or_none()
+        if job is None:
+            raise SoftwareJobError("job not found")
+        if job.status in {"succeeded", "failed", "cancelled"}:
+            return _job_dict(job), None
+        if job.status == "queued" or job.node_id is None or job.lease_id is None:
+            job.status = "cancelled"
+            job.stage = "terminal"
+            job.error = {"code": "USER_CANCELLED", "detail": "Cancelled before dispatch."}
+            job.terminal_at = now
+            return _job_dict(job), None
+        job.status = "cancelling"
+        job.stage = "cancel_requested"
+        payload = {
+            "job_id": str(job.job_id),
+            "lease_id": str(job.lease_id),
+            "request_digest": job.request_digest,
+        }
+        return _job_dict(job), {"node_id": job.node_id, "payload": payload}
+
+
+def pending_node_cancellations(node_id: UUID) -> list[dict]:
+    """节点重连或心跳时重放未确认的取消意图。"""
+    with session_scope() as session:
+        rows = session.execute(
+            select(SoftwareJob).where(
+                SoftwareJob.node_id == node_id,
+                SoftwareJob.status == "cancelling",
+            )
+        ).scalars()
+        return [
+            {
+                "job_id": str(job.job_id),
+                "lease_id": str(job.lease_id),
+                "request_digest": job.request_digest,
+            }
+            for job in rows
+            if job.lease_id is not None
+        ]
+
+
+def create_job(
+    user_id: UUID,
+    task_id: UUID,
+    *,
+    idempotency_key: str,
+    capability: str,
+    request: dict,
+) -> tuple[dict, bool]:
+    key = idempotency_key.strip()
+    if not key or len(key) > 200:
+        raise SoftwareJobError("idempotency_key must contain 1 to 200 characters")
+    if capability not in SUPPORTED_CAPABILITIES:
+        raise SoftwareJobError("unsupported capability")
+    normalized, digest = _canonical_request(request)
+    with session_scope() as session:
+        task = session.execute(
+            select(Task.task_id).where(Task.task_id == task_id, Task.user_id == user_id)
+        ).first()
+        if task is None:
+            raise SoftwareJobError("task not found")
+        artifact_id = UUID(normalized["input"]["input_id"])
+        artifact = session.execute(
+            select(Artifact).where(
+                Artifact.artifact_id == artifact_id,
+                Artifact.user_id == user_id,
+                Artifact.status == "active",
+            )
+        ).scalar_one_or_none()
+        if artifact is None:
+            raise SoftwareJobError("input artifact not found")
+        suffix = "." + artifact.current_path.rsplit(".", 1)[-1].lower() if "." in artifact.current_path else ""
+        if suffix not in ALLOWED_INPUT_SUFFIXES:
+            raise SoftwareJobError("input artifact type is not supported")
+        if (
+            artifact.size_bytes is None
+            or artifact.size_bytes < 0
+            or artifact.size_bytes > MAX_INPUT_BYTES
+            or not artifact.content_sha256
+            or len(artifact.content_sha256) != 64
+        ):
+            raise SoftwareJobError("input artifact metadata is incomplete or too large")
+        input_manifest = {
+            "artifact_id": str(artifact.artifact_id),
+            "filename": artifact.current_path.replace("\\", "/").rsplit("/", 1)[-1],
+            "size_bytes": artifact.size_bytes,
+            "sha256": artifact.content_sha256,
+        }
+        existing = session.execute(
+            select(SoftwareJob).where(
+                SoftwareJob.user_id == user_id,
+                SoftwareJob.idempotency_key == key,
+            )
+        ).scalar_one_or_none()
+        if existing is not None:
+            if (
+                existing.task_id != task_id
+                or existing.capability != capability
+                or existing.request_digest != digest
+            ):
+                raise SoftwareJobError("idempotency key was already used for a different request")
+            return _job_dict(existing), False
+        row = SoftwareJob(
+            job_id=uuid4(),
+            user_id=user_id,
+            task_id=task_id,
+            idempotency_key=key,
+            capability=capability,
+            request=normalized,
+            request_digest=digest,
+            input_manifest=input_manifest,
+            status="queued",
+            stage="",
+            metrics={},
+            error={},
+            artifact_manifest=[],
+        )
+        try:
+            with session.begin_nested():
+                session.add(row)
+                session.flush()
+            return _job_dict(row), True
+        except IntegrityError:
+            existing = session.execute(
+                select(SoftwareJob).where(
+                    SoftwareJob.user_id == user_id,
+                    SoftwareJob.idempotency_key == key,
+                )
+            ).scalar_one()
+            if (
+                existing.task_id != task_id
+                or existing.capability != capability
+                or existing.request_digest != digest
+            ):
+                raise SoftwareJobError(
+                    "idempotency key was already used for a different request"
+                )
+            return _job_dict(existing), False
+
+
+def get_job(user_id: UUID, job_id: UUID) -> dict | None:
+    with session_scope() as session:
+        row = session.execute(
+            select(SoftwareJob).where(SoftwareJob.job_id == job_id, SoftwareJob.user_id == user_id)
+        ).scalar_one_or_none()
+        return _job_dict(row) if row else None
+
+
+def offer_next_job(node_ids: set[UUID]) -> dict | None:
+    """从当前进程实际在线的节点中选择一个，为最早 queued job 创建短租约。"""
+    if not node_ids:
+        return None
+    now = datetime.now(timezone.utc)
+    with session_scope() as session:
+        expired = session.execute(
+            select(SoftwareJob)
+            .where(
+                SoftwareJob.status == "offered",
+                SoftwareJob.lease_expires_at <= now,
+            )
+            .with_for_update(skip_locked=True)
+        ).scalars()
+        for item in expired:
+            item.status = "queued"
+            item.node_id = None
+            item.lease_id = None
+            item.lease_expires_at = None
+        job = session.execute(
+            select(SoftwareJob)
+            .where(SoftwareJob.status == "queued")
+            .order_by(SoftwareJob.created_at, SoftwareJob.job_id)
+            .with_for_update(skip_locked=True)
+            .limit(1)
+        ).scalar_one_or_none()
+        if job is None:
+            return None
+        busy_node_ids = set(
+            session.execute(
+                select(SoftwareJob.node_id).where(
+                    SoftwareJob.node_id.is_not(None),
+                    SoftwareJob.status.in_({"offered", "dispatched", "running"}),
+                )
+            ).scalars()
+        )
+        nodes = session.execute(
+            select(SoftwareNode)
+            .where(SoftwareNode.node_id.in_(node_ids), SoftwareNode.status == "online")
+            .order_by(SoftwareNode.last_seen_at.desc())
+        ).scalars()
+        node = next(
+            (
+                item
+                for item in nodes
+                if item.node_id not in busy_node_ids
+                and job.capability in item.capabilities
+                and int((item.runtime or {}).get("available_slots") or 0) > 0
+            ),
+            None,
+        )
+        if node is None:
+            return None
+        lease_id = uuid4()
+        expires_at = now + timedelta(seconds=OFFER_SECONDS)
+        job.node_id = node.node_id
+        job.lease_id = lease_id
+        job.lease_expires_at = expires_at
+        job.status = "offered"
+        return {
+            "node_id": node.node_id,
+            "payload": {
+                "job_id": str(job.job_id),
+                "lease_id": str(lease_id),
+                "lease_expires_at": expires_at.isoformat(),
+                "capability": job.capability,
+                "request_digest": job.request_digest,
+                "request": job.request,
+                "input_transfer": {
+                    **job.input_manifest,
+                    "download_path": f"/v1/software-jobs/{job.job_id}/input",
+                },
+            },
+        }
+
+
+def get_job_input(node_id: UUID, job_id: UUID) -> dict | None:
+    """返回任务绑定的 artifact 定位信息；调用方仍需在 user_root 内安全解析。"""
+    with session_scope() as session:
+        job = session.execute(
+            select(SoftwareJob).where(
+                SoftwareJob.job_id == job_id,
+                SoftwareJob.node_id == node_id,
+                SoftwareJob.status.in_({"offered", "dispatched", "running", "disconnected"}),
+            )
+        ).scalar_one_or_none()
+        if job is None:
+            return None
+        artifact_id = UUID(job.input_manifest["artifact_id"])
+        artifact = session.execute(
+            select(Artifact).where(
+                Artifact.artifact_id == artifact_id,
+                Artifact.user_id == job.user_id,
+                Artifact.status == "active",
+            )
+        ).scalar_one_or_none()
+        if artifact is None:
+            return None
+        return {
+            "user_id": job.user_id,
+            "current_path": artifact.current_path,
+            **job.input_manifest,
+        }
+
+
+def get_job_output_context(node_id: UUID, job_id: UUID, lease_id: UUID, digest: str) -> dict | None:
+    """返回 Node 输出上传上下文，不向 Node 暴露任何云端文件路径。"""
+    with session_scope() as session:
+        row = session.execute(
+            select(SoftwareJob, Task.working_dir)
+            .join(Task, Task.task_id == SoftwareJob.task_id)
+            .where(SoftwareJob.job_id == job_id)
+        ).one_or_none()
+        if row is None:
+            return None
+        job, working_dir = row
+        if (
+            job.node_id != node_id
+            or job.lease_id != lease_id
+            or job.request_digest != digest
+            or job.status not in {"dispatched", "running", "disconnected", "succeeded"}
+        ):
+            return None
+        return {
+            "user_id": job.user_id,
+            "task_id": job.task_id,
+            "working_dir": working_dir,
+            "request": job.request,
+            "status": job.status,
+            "artifact_manifest": job.artifact_manifest,
+        }
+
+
+def validate_output_manifest(request: dict, manifest: object) -> list[dict]:
+    if not isinstance(manifest, list):
+        raise SoftwareJobError("job artifact manifest must be a list")
+    requested_formats = set(request.get("output", {}).get("formats") or [])
+    expected_ids = {"plot_spec", "provenance"}
+    expected_ids.update(
+        artifact_id
+        for artifact_id, (_, _, output_format) in OUTPUT_ARTIFACTS.items()
+        if output_format in requested_formats
+    )
+    if len(manifest) != len(expected_ids):
+        raise SoftwareJobError("job artifact manifest is incomplete")
+    normalized: list[dict] = []
+    seen: set[str] = set()
+    total = 0
+    for raw in manifest:
+        if not isinstance(raw, dict) or set(raw) != {
+            "artifact_id", "filename", "media_type", "size_bytes", "sha256"
+        }:
+            raise SoftwareJobError("job artifact manifest entry is invalid")
+        local_id = raw.get("artifact_id")
+        if local_id not in expected_ids or local_id in seen:
+            raise SoftwareJobError("job artifact manifest identity is invalid")
+        filename, media_type, _ = OUTPUT_ARTIFACTS[local_id]
+        size = raw.get("size_bytes")
+        digest = raw.get("sha256")
+        if raw.get("filename") != filename or raw.get("media_type") != media_type:
+            raise SoftwareJobError("job artifact manifest metadata does not match its identity")
+        if not isinstance(size, int) or isinstance(size, bool) or not 1 <= size <= MAX_OUTPUT_ARTIFACT_BYTES:
+            raise SoftwareJobError("job output artifact size is invalid")
+        if not isinstance(digest, str) or not re.fullmatch(r"[0-9a-f]{64}", digest):
+            raise SoftwareJobError("job output artifact digest is invalid")
+        total += size
+        seen.add(local_id)
+        normalized.append(dict(raw))
+    if seen != expected_ids or total > MAX_OUTPUT_TOTAL_BYTES:
+        raise SoftwareJobError("job artifact manifest is incomplete or too large")
+    return normalized
+
+
+def abandon_offer(node_id: UUID, payload: dict) -> None:
+    """WebSocket 发送失败时只回滚仍属于该连接租约的 offer。"""
+    try:
+        job_id = UUID(str(payload.get("job_id", "")))
+        lease_id = UUID(str(payload.get("lease_id", "")))
+    except ValueError:
+        return
+    with session_scope() as session:
+        job = session.execute(
+            select(SoftwareJob).where(SoftwareJob.job_id == job_id).with_for_update()
+        ).scalar_one_or_none()
+        if (
+            job is not None
+            and job.status == "offered"
+            and job.node_id == node_id
+            and job.lease_id == lease_id
+        ):
+            job.status = "queued"
+            job.node_id = None
+            job.lease_id = None
+            job.lease_expires_at = None
+
+
+def respond_to_offer(node_id: UUID, *, accepted: bool, payload: dict) -> None:
+    try:
+        job_id = UUID(str(payload.get("job_id", "")))
+        lease_id = UUID(str(payload.get("lease_id", "")))
+    except ValueError as exc:
+        raise SoftwareJobError("invalid job offer response identity") from exc
+    now = datetime.now(timezone.utc)
+    with session_scope() as session:
+        job = session.execute(
+            select(SoftwareJob).where(SoftwareJob.job_id == job_id).with_for_update()
+        ).scalar_one_or_none()
+        if job is None or job.node_id != node_id or job.lease_id != lease_id:
+            raise SoftwareJobError("job offer is stale or does not belong to this node")
+        if (
+            accepted
+            and job.status in {"dispatched", "running", "cancelling", "succeeded", "failed", "cancelled"}
+            and payload.get("request_digest") == job.request_digest
+        ):
+            return
+        if job.status != "offered":
+            raise SoftwareJobError("job offer is stale or does not belong to this node")
+        if job.lease_expires_at is None or job.lease_expires_at <= now:
+            job.status = "queued"
+            job.node_id = None
+            job.lease_id = None
+            job.lease_expires_at = None
+            raise SoftwareJobError("job offer has expired")
+        if accepted:
+            if payload.get("request_digest") != job.request_digest:
+                raise SoftwareJobError("job request digest mismatch")
+            job.status = "dispatched"
+            job.stage = "accepted"
+            job.error = {}
+        else:
+            job.status = "queued"
+            job.node_id = None
+            job.lease_id = None
+            job.lease_expires_at = None
+            job.error = {"code": "node_rejected", "detail": str(payload.get("reason") or "")[:500]}
+
+
+def update_job_state(node_id: UUID, payload: dict) -> None:
+    job_id, lease_id, digest = _message_identity(payload)
+    stage = str(payload.get("stage") or "")
+    progress = payload.get("progress")
+    metrics = payload.get("metrics") or {}
+    if not stage or len(stage) > 100:
+        raise SoftwareJobError("job stage is required")
+    if not isinstance(progress, int) or isinstance(progress, bool) or not 0 <= progress <= 100:
+        raise SoftwareJobError("job progress must be between 0 and 100")
+    if not isinstance(metrics, dict) or len(json.dumps(metrics, ensure_ascii=False)) > 64 * 1024:
+        raise SoftwareJobError("job metrics are invalid")
+    now = datetime.now(timezone.utc)
+    with session_scope() as session:
+        job = session.execute(
+            select(SoftwareJob).where(SoftwareJob.job_id == job_id).with_for_update()
+        ).scalar_one_or_none()
+        _assert_job_message(job, node_id, lease_id, digest)
+        if job.status in {"succeeded", "failed", "cancelled", "cancelling"}:
+            return
+        if not _can_accept_state(job.status):
+            raise SoftwareJobError("job state cannot advance from its current status")
+        job.status = (
+            "dispatched"
+            if stage in {"accepted", "waiting_input", "ready_to_run"}
+            else "running"
+        )
+        job.stage = stage
+        job.progress = progress
+        job.metrics = metrics
+        if job.status == "running" and job.started_at is None:
+            job.started_at = now
+
+
+def record_job_terminal(node_id: UUID, payload: dict) -> None:
+    job_id, lease_id, digest = _message_identity(payload)
+    terminal_status = payload.get("status")
+    if terminal_status not in {"succeeded", "failed", "cancelled"}:
+        raise SoftwareJobError("invalid job terminal status")
+    error = payload.get("error") or {}
+    manifest = payload.get("artifact_manifest") or []
+    if not isinstance(error, dict) or len(json.dumps(error, ensure_ascii=False)) > 64 * 1024:
+        raise SoftwareJobError("job terminal error is invalid")
+    if not isinstance(manifest, list) or len(json.dumps(manifest, ensure_ascii=False)) > 256 * 1024:
+        raise SoftwareJobError("job artifact manifest is invalid")
+    now = datetime.now(timezone.utc)
+    with session_scope() as session:
+        job = session.execute(
+            select(SoftwareJob).where(SoftwareJob.job_id == job_id).with_for_update()
+        ).scalar_one_or_none()
+        _assert_job_message(job, node_id, lease_id, digest)
+        if terminal_status == "succeeded":
+            expected = validate_output_manifest(
+                job.request,
+                [
+                    {
+                        "artifact_id": item.get("source_artifact_id"),
+                        "filename": item.get("filename"),
+                        "media_type": item.get("media_type"),
+                        "size_bytes": item.get("size_bytes"),
+                        "sha256": item.get("sha256"),
+                    }
+                    for item in manifest
+                    if isinstance(item, dict)
+                ],
+            )
+            if len(expected) != len(manifest) or any(
+                not isinstance(item.get("artifact_id"), str)
+                or not _is_uuid(item["artifact_id"])
+                or not isinstance(item.get("path"), str)
+                or not item["path"].startswith(f"origin/{job.job_id}/")
+                for item in manifest
+            ):
+                raise SoftwareJobError("successful job artifacts have not been published")
+        if job.status in {"succeeded", "failed", "cancelled"}:
+            if job.status != terminal_status:
+                raise SoftwareJobError("job terminal status conflicts with existing terminal")
+            return
+        if job.status not in {"offered", "dispatched", "running", "disconnected", "cancelling"}:
+            raise SoftwareJobError("job terminal cannot advance from its current status")
+        job.status = terminal_status
+        job.stage = "terminal"
+        job.progress = 100 if terminal_status == "succeeded" else job.progress
+        job.error = error
+        job.artifact_manifest = manifest
+        job.terminal_at = now
+
+
+def _is_uuid(value: str) -> bool:
+    try:
+        UUID(value)
+    except ValueError:
+        return False
+    return True
+
+
+def mark_node_jobs_disconnected(node_id: UUID) -> None:
+    """连接丢失后保留 Node 归属和 lease，禁止任务被自动重派。"""
+    with session_scope() as session:
+        jobs = session.execute(
+            select(SoftwareJob)
+            .where(
+                SoftwareJob.node_id == node_id,
+                SoftwareJob.status.in_({"dispatched", "running"}),
+            )
+            .with_for_update()
+        ).scalars()
+        for job in jobs:
+            job.status = "disconnected"
+
+
+def _can_accept_state(status: str) -> bool:
+    return status in {"offered", "dispatched", "running", "disconnected"}
+
+
+def _message_identity(payload: dict) -> tuple[UUID, UUID, str]:
+    try:
+        job_id = UUID(str(payload.get("job_id", "")))
+        lease_id = UUID(str(payload.get("lease_id", "")))
+    except ValueError as exc:
+        raise SoftwareJobError("invalid job message identity") from exc
+    digest = str(payload.get("request_digest") or "")
+    if len(digest) != 64:
+        raise SoftwareJobError("invalid job request digest")
+    return job_id, lease_id, digest
+
+
+def _assert_job_message(
+    job: SoftwareJob | None,
+    node_id: UUID,
+    lease_id: UUID,
+    digest: str,
+) -> None:
+    if (
+        job is None
+        or job.node_id != node_id
+        or job.lease_id != lease_id
+        or job.request_digest != digest
+    ):
+        raise SoftwareJobError("job message does not belong to this node or lease")

core/software_nodes.py

@@ -11,13 +11,13 @@ import bcrypt
 from sqlalchemy import select
 
 from core.storage.engine import session_scope
-from core.storage.models import ComputeNode, ComputeNodeEnrollment
+from core.storage.models import SoftwareNode, SoftwareNodeEnrollment
 
 SUPPORTED_CAPABILITIES = frozenset({"origin.plot@v1"})
 MAX_ENROLLMENT_FAILURES = 5
 
 
-class ComputeNodeError(Exception):
+class SoftwareNodeError(Exception):
     pass
 
 
@@ -46,12 +46,12 @@ def create_enrollment(
 ) -> dict:
     allowed = list(dict.fromkeys(capabilities or ["origin.plot@v1"]))
     if not allowed or any(item not in SUPPORTED_CAPABILITIES for item in allowed):
-        raise ComputeNodeError("unsupported capability")
+        raise SoftwareNodeError("unsupported capability")
     if not 60 <= ttl_seconds <= 3600:
-        raise ComputeNodeError("ttl_seconds must be between 60 and 3600")
+        raise SoftwareNodeError("ttl_seconds must be between 60 and 3600")
     code = "ZCN-" + secrets.token_hex(16).upper()
     expires_at = datetime.now(timezone.utc) + timedelta(seconds=ttl_seconds)
-    row = ComputeNodeEnrollment(
+    row = SoftwareNodeEnrollment(
         enrollment_id=uuid4(),
         code_hash=_enrollment_digest(code),
         expected_name=(expected_name or "").strip() or None,
@@ -80,22 +80,22 @@ def enroll_node(
     name = node_name.strip()
     requested = list(dict.fromkeys(capabilities))
     if not name or not requested:
-        raise ComputeNodeError("node_name and capabilities are required")
+        raise SoftwareNodeError("node_name and capabilities are required")
     now = datetime.now(timezone.utc)
     error: str | None = None
     with session_scope() as session:
         enrollment = session.execute(
-            select(ComputeNodeEnrollment)
+            select(SoftwareNodeEnrollment)
             .where(
-                ComputeNodeEnrollment.code_hash == _enrollment_digest(enrollment_code),
-                ComputeNodeEnrollment.consumed_at.is_(None),
-                ComputeNodeEnrollment.expires_at > now,
-                ComputeNodeEnrollment.failed_attempts < MAX_ENROLLMENT_FAILURES,
+                SoftwareNodeEnrollment.code_hash == _enrollment_digest(enrollment_code),
+                SoftwareNodeEnrollment.consumed_at.is_(None),
+                SoftwareNodeEnrollment.expires_at > now,
+                SoftwareNodeEnrollment.failed_attempts < MAX_ENROLLMENT_FAILURES,
             )
             .with_for_update()
         ).scalar_one_or_none()
         if enrollment is None:
-            raise ComputeNodeError("invalid or expired enrollment code")
+            raise SoftwareNodeError("invalid or expired enrollment code")
         enrollment.failed_attempts += 1
         if enrollment.expected_name and enrollment.expected_name != name:
             error = "node name does not match enrollment"
@@ -103,7 +103,7 @@ def enroll_node(
             error = "capability is not allowed by enrollment"
         else:
             existing = session.execute(
-                select(ComputeNode.node_id).where(ComputeNode.install_id == install_id)
+                select(SoftwareNode.node_id).where(SoftwareNode.install_id == install_id)
             ).first()
             if existing is not None:
                 error = "install is already enrolled"
@@ -111,7 +111,7 @@ def enroll_node(
                 token = secrets.token_urlsafe(48)
                 node_id = uuid4()
                 session.add(
-                    ComputeNode(
+                    SoftwareNode(
                         node_id=node_id,
                         name=name,
                         install_id=install_id,
@@ -125,7 +125,7 @@ def enroll_node(
                 )
                 enrollment.consumed_at = now
     if error is not None:
-        raise ComputeNodeError(error)
+        raise SoftwareNodeError(error)
     return {
         "node_id": str(node_id),
         "node_token": token,
@@ -136,13 +136,13 @@ def enroll_node(
 
 def authenticate_node(node_id: UUID, token: str) -> dict:
     with session_scope() as session:
-        node = session.get(ComputeNode, node_id)
+        node = session.get(SoftwareNode, node_id)
         if (
             node is None
             or node.status == "disabled"
             or not _verify_secret(token, node.token_hash)
         ):
-            raise ComputeNodeError("invalid node credentials")
+            raise SoftwareNodeError("invalid node credentials")
         return {
             "node_id": node.node_id,
             "install_id": node.install_id,
@@ -152,9 +152,9 @@ def authenticate_node(node_id: UUID, token: str) -> dict:
 
 def update_node_runtime(node_id: UUID, *, status: str, runtime: dict) -> None:
     with session_scope() as session:
-        node = session.get(ComputeNode, node_id)
+        node = session.get(SoftwareNode, node_id)
         if node is None or node.status == "disabled":
-            raise ComputeNodeError("node is disabled or missing")
+            raise SoftwareNodeError("node is disabled or missing")
         node.status = status
         node.runtime = runtime
         node.last_seen_at = datetime.now(timezone.utc)
@@ -163,14 +163,14 @@ def update_node_runtime(node_id: UUID, *, status: str, runtime: dict) -> None:
 def mark_node_offline(node_id: UUID) -> None:
     """仅把活动节点转离线；管理员禁用态不可被断线收尾覆盖。"""
     with session_scope() as session:
-        node = session.get(ComputeNode, node_id)
+        node = session.get(SoftwareNode, node_id)
         if node is not None and node.status != "disabled":
             node.status = "offline"
 
 
 def set_node_disabled(node_id: UUID, disabled: bool) -> bool:
     with session_scope() as session:
-        node = session.get(ComputeNode, node_id)
+        node = session.get(SoftwareNode, node_id)
         if node is None:
             return False
         node.status = "disabled" if disabled else "offline"
@@ -180,7 +180,7 @@ def set_node_disabled(node_id: UUID, disabled: bool) -> bool:
 def delete_node(node_id: UUID) -> bool:
     """撤销并物理删除节点身份；当前节点表没有任务历史外键。"""
     with session_scope() as session:
-        node = session.get(ComputeNode, node_id)
+        node = session.get(SoftwareNode, node_id)
         if node is None:
             return False
         session.delete(node)
@@ -190,7 +190,7 @@ def delete_node(node_id: UUID) -> bool:
 def list_nodes() -> list[dict]:
     with session_scope() as session:
         rows = (
-            session.execute(select(ComputeNode).order_by(ComputeNode.created_at))
+            session.execute(select(SoftwareNode).order_by(SoftwareNode.created_at))
             .scalars()
             .all()
         )

core/storage/models.py

@@ -424,10 +424,10 @@ class ChannelBinding(Base):
     )
 
 
-class ComputeNodeEnrollment(Base):
+class SoftwareNodeEnrollment(Base):
     """Windows Node 一次性注册码；数据库只保存不可逆摘要。"""
 
-    __tablename__ = "compute_node_enrollments"
+    __tablename__ = "software_node_enrollments"
     enrollment_id: Mapped[UUID] = mapped_column(PG_UUID(as_uuid=True), primary_key=True, default=uuid4)
     code_hash: Mapped[str] = mapped_column(Text, nullable=False, unique=True)
     expected_name: Mapped[Optional[str]] = mapped_column(Text, nullable=True)
@@ -443,10 +443,10 @@ class ComputeNodeEnrollment(Base):
     )
 
 
-class ComputeNode(Base):
+class SoftwareNode(Base):
     """平台托管的 Windows 执行节点身份与最后一次运行态。"""
 
-    __tablename__ = "compute_nodes"
+    __tablename__ = "software_nodes"
     node_id: Mapped[UUID] = mapped_column(PG_UUID(as_uuid=True), primary_key=True, default=uuid4)
     name: Mapped[str] = mapped_column(Text, nullable=False)
     install_id: Mapped[UUID] = mapped_column(PG_UUID(as_uuid=True), nullable=False, unique=True)
@@ -465,6 +465,49 @@ class ComputeNode(Base):
     )
 
 
+class SoftwareJob(Base):
+    """专业软件任务账本；请求只保存规范化参数和输入引用。"""
+
+    __tablename__ = "software_jobs"
+    __table_args__ = (
+        UniqueConstraint("user_id", "idempotency_key", name="uq_software_jobs_user_idempotency"),
+        Index("ix_software_jobs_status_created", "status", "created_at"),
+        Index("ix_software_jobs_node_status", "node_id", "status"),
+    )
+
+    job_id: Mapped[UUID] = mapped_column(PG_UUID(as_uuid=True), primary_key=True, default=uuid4)
+    user_id: Mapped[UUID] = mapped_column(
+        PG_UUID(as_uuid=True), ForeignKey("users.user_id", ondelete="CASCADE"), nullable=False
+    )
+    task_id: Mapped[UUID] = mapped_column(
+        PG_UUID(as_uuid=True), ForeignKey("tasks.task_id", ondelete="CASCADE"), nullable=False
+    )
+    idempotency_key: Mapped[str] = mapped_column(Text, nullable=False)
+    capability: Mapped[str] = mapped_column(Text, nullable=False)
+    request: Mapped[dict[str, Any]] = mapped_column(JSONB, nullable=False)
+    request_digest: Mapped[str] = mapped_column(Text, nullable=False)
+    input_manifest: Mapped[dict[str, Any]] = mapped_column(JSONB, nullable=False, default=dict)
+    node_id: Mapped[Optional[UUID]] = mapped_column(
+        PG_UUID(as_uuid=True), ForeignKey("software_nodes.node_id", ondelete="SET NULL"), nullable=True
+    )
+    lease_id: Mapped[Optional[UUID]] = mapped_column(PG_UUID(as_uuid=True), nullable=True)
+    lease_expires_at: Mapped[Optional[datetime]] = mapped_column(DateTime(timezone=True), nullable=True)
+    status: Mapped[str] = mapped_column(Text, nullable=False, default="queued", server_default="queued")
+    stage: Mapped[str] = mapped_column(Text, nullable=False, default="", server_default="")
+    progress: Mapped[int] = mapped_column(Integer, nullable=False, default=0, server_default="0")
+    metrics: Mapped[dict[str, Any]] = mapped_column(JSONB, nullable=False, default=dict)
+    error: Mapped[dict[str, Any]] = mapped_column(JSONB, nullable=False, default=dict)
+    artifact_manifest: Mapped[list[Any]] = mapped_column(JSONB, nullable=False, default=list)
+    started_at: Mapped[Optional[datetime]] = mapped_column(DateTime(timezone=True), nullable=True)
+    terminal_at: Mapped[Optional[datetime]] = mapped_column(DateTime(timezone=True), nullable=True)
+    created_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True), server_default=func.now(), nullable=False
+    )
+    updated_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True), server_default=func.now(), onupdate=func.now(), nullable=False
+    )
+
+
 class ExternalSystemDefinition(Base):
     """管理员维护的可信外部系统目录；不含任何用户凭据。"""
 

core/tool_registry.py

@@ -47,6 +47,12 @@ from tools.run_python import RunPythonTool
 from tools.schedule import (
     ScheduleCancelTool, ScheduleCreateTool, ScheduleListTool, ScheduleUpdateTool,
 )
+from tools.software_jobs import (
+    SoftwareCapabilityListTool,
+    SoftwareJobCancelTool,
+    SoftwareJobStatusTool,
+    SoftwareJobSubmitTool,
+)
 from tools.seedance import SeedanceTool
 from tools.seedream import SeedreamTool
 from tools.send_email import SendEmailTool, smtp_configured
@@ -207,6 +213,14 @@ def build_tools(ctx: ToolContext) -> dict[str, Any]:
             ScheduleCancelTool(ctx.uid, **base),
         ]
 
+    def _software_jobs() -> list:
+        return [
+            SoftwareCapabilityListTool(ctx.uid, ctx.task_id, **base),
+            SoftwareJobSubmitTool(ctx.uid, ctx.task_id, **base),
+            SoftwareJobStatusTool(ctx.uid, ctx.task_id, **base),
+            SoftwareJobCancelTool(ctx.uid, ctx.task_id, **base),
+        ]
+
     def _send_email() -> list:
         # base_dir 用 working_dir_path(宿主 task 目录):send_email 在宿主进程读附件,
         # docker 下 agent 给的相对路径相对容器 workdir=task_dir,翻回宿主即 working_dir_path;
@@ -294,6 +308,7 @@ def build_tools(ctx: ToolContext) -> dict[str, Any]:
         ("skill_authoring",  lambda: True,                          _skill_authoring),
         # 定时 run 内不挂 schedule_*(防任务造任务自我繁殖);仅交互对话可建/管 job。
         ("schedules",        lambda: not ctx.scheduled_run,         _schedules),
+        ("software_jobs",    lambda: not ctx.scheduled_run,         _software_jobs),
         ("send_email",       smtp_configured,                       _send_email),
         ("wechat_push",      wechat_push_available,                 _wechat_push),
         ("run_python",       lambda: ctx.caps.enable_run_python,    _run_python),

db/migrations/versions/20260813_1600_0032_software_jobs.py

@@ -0,0 +1,102 @@
+"""Rename software nodes and add the professional software job ledger.
+
+Revision ID: 0032
+Revises: 0031
+Create Date: 2026-08-13
+"""
+from collections.abc import Sequence
+
+import sqlalchemy as sa
+from alembic import op
+from sqlalchemy.dialects import postgresql
+
+revision: str = "0032"
+down_revision: str | None = "0031"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def upgrade() -> None:
+    op.rename_table("compute_node_enrollments", "software_node_enrollments")
+    op.rename_table("compute_nodes", "software_nodes")
+    op.execute(
+        "ALTER INDEX ix_compute_nodes_status RENAME TO ix_software_nodes_status"
+    )
+    op.execute(
+        "ALTER TABLE software_node_enrollments RENAME CONSTRAINT "
+        "compute_node_enrollments_pkey TO software_node_enrollments_pkey"
+    )
+    op.execute(
+        "ALTER TABLE software_node_enrollments RENAME CONSTRAINT "
+        "compute_node_enrollments_code_hash_key TO software_node_enrollments_code_hash_key"
+    )
+    op.execute(
+        "ALTER TABLE software_node_enrollments RENAME CONSTRAINT "
+        "compute_node_enrollments_created_by_fkey TO software_node_enrollments_created_by_fkey"
+    )
+    op.execute(
+        "ALTER TABLE software_nodes RENAME CONSTRAINT "
+        "compute_nodes_pkey TO software_nodes_pkey"
+    )
+    op.execute(
+        "ALTER TABLE software_nodes RENAME CONSTRAINT "
+        "compute_nodes_install_id_key TO software_nodes_install_id_key"
+    )
+    op.create_table(
+        "software_jobs",
+        sa.Column("job_id", postgresql.UUID(as_uuid=True), primary_key=True),
+        sa.Column("user_id", postgresql.UUID(as_uuid=True), sa.ForeignKey("users.user_id", ondelete="CASCADE"), nullable=False),
+        sa.Column("task_id", postgresql.UUID(as_uuid=True), sa.ForeignKey("tasks.task_id", ondelete="CASCADE"), nullable=False),
+        sa.Column("idempotency_key", sa.Text(), nullable=False),
+        sa.Column("capability", sa.Text(), nullable=False),
+        sa.Column("request", postgresql.JSONB(), nullable=False),
+        sa.Column("request_digest", sa.Text(), nullable=False),
+        sa.Column("input_manifest", postgresql.JSONB(), nullable=False),
+        sa.Column("node_id", postgresql.UUID(as_uuid=True), sa.ForeignKey("software_nodes.node_id", ondelete="SET NULL"), nullable=True),
+        sa.Column("lease_id", postgresql.UUID(as_uuid=True), nullable=True),
+        sa.Column("lease_expires_at", sa.DateTime(timezone=True), nullable=True),
+        sa.Column("status", sa.Text(), server_default="queued", nullable=False),
+        sa.Column("stage", sa.Text(), server_default="", nullable=False),
+        sa.Column("progress", sa.Integer(), server_default="0", nullable=False),
+        sa.Column("metrics", postgresql.JSONB(), nullable=False),
+        sa.Column("error", postgresql.JSONB(), nullable=False),
+        sa.Column("artifact_manifest", postgresql.JSONB(), nullable=False),
+        sa.Column("started_at", sa.DateTime(timezone=True), nullable=True),
+        sa.Column("terminal_at", sa.DateTime(timezone=True), nullable=True),
+        sa.Column("created_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
+        sa.Column("updated_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
+        sa.UniqueConstraint("user_id", "idempotency_key", name="uq_software_jobs_user_idempotency"),
+    )
+    op.create_index("ix_software_jobs_status_created", "software_jobs", ["status", "created_at"])
+    op.create_index("ix_software_jobs_node_status", "software_jobs", ["node_id", "status"])
+
+
+def downgrade() -> None:
+    op.drop_index("ix_software_jobs_node_status", table_name="software_jobs")
+    op.drop_index("ix_software_jobs_status_created", table_name="software_jobs")
+    op.drop_table("software_jobs")
+    op.execute(
+        "ALTER TABLE software_nodes RENAME CONSTRAINT "
+        "software_nodes_install_id_key TO compute_nodes_install_id_key"
+    )
+    op.execute(
+        "ALTER TABLE software_nodes RENAME CONSTRAINT "
+        "software_nodes_pkey TO compute_nodes_pkey"
+    )
+    op.execute(
+        "ALTER TABLE software_node_enrollments RENAME CONSTRAINT "
+        "software_node_enrollments_created_by_fkey TO compute_node_enrollments_created_by_fkey"
+    )
+    op.execute(
+        "ALTER TABLE software_node_enrollments RENAME CONSTRAINT "
+        "software_node_enrollments_code_hash_key TO compute_node_enrollments_code_hash_key"
+    )
+    op.execute(
+        "ALTER TABLE software_node_enrollments RENAME CONSTRAINT "
+        "software_node_enrollments_pkey TO compute_node_enrollments_pkey"
+    )
+    op.execute(
+        "ALTER INDEX ix_software_nodes_status RENAME TO ix_compute_nodes_status"
+    )
+    op.rename_table("software_nodes", "compute_nodes")
+    op.rename_table("software_node_enrollments", "compute_node_enrollments")

deploy/nginx/zcbot.conf.example

@@ -60,7 +60,7 @@ server {
 
     # ★ Windows Node 长连接:注册走普通 POST,注册后的节点控制通道走 WebSocket。
     # 必须独立于下面会清空 Connection 头的默认 location；15s 应用心跳保持链路活跃。
-    location = /v1/compute/nodes/connect {
+    location = /v1/software-nodes/connect {
         proxy_pass http://zcbot_backend;
         proxy_http_version 1.1;
         proxy_set_header Upgrade $http_upgrade;

docs/windows-node-design.md

@@ -34,7 +34,7 @@ Windows Node 不是完整的本地 zcbot：
 ### 2.1 目标
 
 - 云端 zcbot 可以发现节点能力、容量、软件版本和在线状态。
-- 用户可以提交、查询、取消长时间 Windows 计算任务。
+- 用户可以提交、查询、取消长时间运行的专业软件任务。
 - 网络中断、云端重启或节点重启后，任务可以确定性对账。
 - 节点可以上报阶段、进度、结构化指标、日志摘要和事件截图。
 - 中间产物和最终产物支持校验、断点上传与按需导入工作目录。
@@ -79,7 +79,7 @@ flowchart LR
 |---|---|
 | `NodeRegistry` | 节点注册、证书指纹、启停、能力和管理员标签 |
 | `NodeConnectionManager` | WSS 连接、心跳、消息 ACK、同节点单活连接 |
-| `ComputeJobService` | 用户授权、幂等提交、节点选择、租约、取消和终态 |
+| `SoftwareJobService` | 用户授权、幂等提交、节点选择、租约、取消和终态 |
 | `ComputeTransferService` | 输入下载凭证、分块上传、SHA-256、容量与保留期 |
 | `ComputeBroker` | 把任务事件推送到 Web UI；不承担持久化事实源 |
 | `ComputeTools` | agent 可调用的能力发现、提交、查询、取消、产物导入工具 |
@@ -157,7 +157,7 @@ sequenceDiagram
     participant Node as Windows Node
     Admin->>Cloud: 创建一次性 enrollment token
     Node->>Node: 生成设备密钥对
-    Node->>Cloud: POST /v1/compute/nodes/enroll
+    Node->>Cloud: POST /v1/software-nodes/enroll
     Cloud->>Cloud: 消耗 token，创建 node_id
     Cloud-->>Node: 客户端证书、CA、云端地址
     Node->>Node: 私钥写入 Windows Certificate Store
@@ -178,7 +178,7 @@ sequenceDiagram
 节点连接：
 
 ```text
-WSS /v1/compute/nodes/connect
+WSS /v1/software-nodes/connect
 ```
 
 统一消息 envelope：
@@ -252,13 +252,13 @@ Node 在 `hello` 和心跳中声明由本机可信配置生成的能力：
 建议新增三张表，不复用外部系统连接表：
 
 ```text
-compute_nodes(
+software_nodes(
   node_id pk, name, cert_fingerprint, status,
   labels jsonb, capabilities jsonb, config_revision,
   last_seen_at, disabled_at, created_at, updated_at
 )
 
-compute_jobs(
+software_jobs(
   job_id pk, user_id fk, task_id fk, tool_call_id,
   capability, schema_version, request jsonb,
   idempotency_key, request_digest,
@@ -268,7 +268,7 @@ compute_jobs(
   created_at, started_at, terminal_at, updated_at
 )
 
-compute_job_events(
+software_job_events(
   event_id pk, job_id fk, sequence,
   kind, level, payload jsonb, created_at
 )
@@ -352,7 +352,7 @@ Node 不访问整个用户 workspace。云端只为显式引用的文件创建
 大产物不通过 WSS 消息传输。Node 使用 HTTPS 分块上传：
 
 ```text
-POST /v1/compute/jobs/{job_id}/artifacts/upload-session
+POST /v1/software-jobs/{job_id}/artifacts/upload-session
 PUT  /v1/compute/transfers/{transfer_id}/parts/{part_number}
 POST /v1/compute/transfers/{transfer_id}/complete
 ```
@@ -364,7 +364,7 @@ POST /v1/compute/transfers/{transfer_id}/complete
 Node 上传完成后先进入：
 
 ```text
-<user_root>/.zcbot_cache/<task_id>/compute_jobs/<job_id>/
+<user_root>/.zcbot_cache/<task_id>/software_jobs/<job_id>/
 ```
 
 该目录默认隐藏且有 TTL。用户或 agent 明确导入后复制到：
@@ -379,10 +379,10 @@ Node 上传完成后先进入：
 
 ```text
 compute_capability_list
-compute_job_submit
-compute_job_status
-compute_job_cancel
-compute_job_artifact_import
+software_job_submit
+software_job_status
+software_job_cancel
+software_job_artifact_import
 ```
 
 - capability list 只返回用户有权使用且有健康节点承载的能力；
@@ -418,7 +418,7 @@ process.cancel_adapter
 - 优先捕获目标窗口；
 - 敏感信息上传前遮罩；
 - 视频默认关闭，显式启用时建议 1280×720、5–10 FPS、H.264 分段；
-- 采集失败不得使计算任务失败；
+- 采集失败不得使专业软件任务失败；
 - 模型只按需读取关键帧，不持续消费完整视频。
 
 ## 12. Origin 首批适配器

docs/windows-node-mvp-intranet.md

@@ -60,7 +60,7 @@ sequenceDiagram
     participant N as Windows Node
     A->>Z: 创建一次性注册码
     A->>N: 输入内网地址和注册码
-    N->>Z: HTTP POST /v1/compute/nodes/enroll
+    N->>Z: HTTP POST /v1/software-nodes/enroll
     Z->>Z: 校验并原子消费注册码
     Z-->>N: node_id + node_token + 配置
     N->>N: DPAPI 加密保存 node_token
@@ -71,7 +71,7 @@ sequenceDiagram
 注册请求：
 
 ```http
-POST http://zcbot.internal:8765/v1/compute/nodes/enroll
+POST http://zcbot.internal:8765/v1/software-nodes/enroll
 Content-Type: application/json
 ```
 
@@ -116,7 +116,7 @@ Content-Type: application/json
 ### 4.1 WS 连接
 
 ```http
-GET ws://zcbot.internal:8765/v1/compute/nodes/connect
+GET ws://zcbot.internal:8765/v1/software-nodes/connect
 Authorization: Bearer <node_token>
 X-Node-Id: <node_id>
 Upgrade: websocket
@@ -170,13 +170,13 @@ RDP：不向公网开放，使用 VPN、堡垒机或云安全登录
 云端首期只增加：
 
 ```text
-compute_nodes(
+software_nodes(
   node_id pk, name, install_id, token_hash, status,
   capabilities jsonb, last_seen_at,
   created_at, updated_at
 )
 
-compute_jobs(
+software_jobs(
   job_id pk, user_id fk, task_id fk,
   idempotency_key, capability, request jsonb,
   node_id fk, status, progress,
@@ -200,6 +200,8 @@ Node 断线且本地任务可能仍在执行时标记 `disconnected`，不得自
 
 ## 7. Origin 任务闭环
 
+当前实现进度：云端任务账本、幂等提交、短期 offer、Node 本地原子保存与 accept/reject 已落地。输入以任务绑定的 artifact UUID 下载，Node 流式校验大小和 SHA-256 后原子保存。固定 Worker 使用管理员安装的隔离 Python 运行时与随程序发布的 `worker.py` 驱动 Origin，生成 OPJU、PNG、SVG、PDF、plot spec、provenance 和原子 `terminal.json`；运行不绑定单次 WebSocket，断线后继续执行。同一进程按 job 去重，Node 重启后不重复启动已留启动标记但无可信终态的任务。成功产物逐项流式上传到云端隐藏暂存区，云端复核任务身份、固定文件名、大小和 SHA-256 后，一次性发布到 `<working_dir>/origin/<job_id>/` 并登记平台 artifact UUID；Node 以 `upload-complete.json` 恢复中断上传。
+
 ```text
 用户上传 CSV/XLSX
 → zcbot 生成受控 plot spec

tests/test_changelog.py

@@ -8,6 +8,10 @@ SAMPLE = """# 更新日志
 
 > 导语在首个条目前,应被丢弃。
 
+## Unreleased
+
+- 尚未发布,不能出现在用户版接口中
+
 ## 0.55.0 — 2026-07-09
 
 - 条目 A
@@ -36,6 +40,12 @@ class ChangelogParseTests(unittest.TestCase):
     def test_preamble_before_first_entry_is_dropped(self) -> None:
         es = _parse_changelog(SAMPLE)
         self.assertNotIn("导语", es[0]["body_md"])
+        self.assertNotIn("尚未发布", es[0]["body_md"])
+
+    def test_unreleased_section_is_not_returned(self) -> None:
+        es = _parse_changelog(SAMPLE)
+        self.assertNotIn("Unreleased", [e["version"] for e in es])
+        self.assertEqual([e["version"] for e in es], ["0.55.0", "0.54.1", "0.53.0"])
 
     def test_real_changelog_file_parses(self) -> None:
         # 仓库里的真文件:格式坏了(条目头打错 / 空正文)在 CI 就翻车,不等用户点开才发现

tests/test_compute_nodes.py

@@ -1,109 +0,0 @@
-from __future__ import annotations
-
-import importlib
-import unittest
-from unittest.mock import AsyncMock, patch
-from uuid import uuid4
-
-from alembic.migration import MigrationContext
-from alembic.operations import Operations
-from sqlalchemy import create_mock_engine
-from sqlalchemy.dialects import postgresql
-
-from core.compute_nodes import (
-    _enrollment_digest,
-    _hash_secret,
-    _verify_secret,
-    delete_node,
-)
-from web.routers.compute_nodes import NodeConnectionManager, _bearer
-
-
-class ComputeNodeSecurityTests(unittest.TestCase):
-    def test_secret_hash_is_salted_and_verifiable(self) -> None:
-        first = _hash_secret("node-secret")
-        second = _hash_secret("node-secret")
-        self.assertNotEqual(first, second)
-        self.assertNotIn("node-secret", first)
-        self.assertTrue(_verify_secret("node-secret", first))
-        self.assertFalse(_verify_secret("wrong", first))
-
-    def test_bearer_parser_rejects_query_style_or_missing_token(self) -> None:
-        self.assertEqual(_bearer("Bearer abc"), "abc")
-        with self.assertRaisesRegex(Exception, "missing node bearer token"):
-            _bearer(None)
-
-    def test_enrollment_digest_does_not_store_plaintext(self) -> None:
-        digest = _enrollment_digest("ZCN-ABC")
-        self.assertEqual(len(digest), 64)
-        self.assertNotIn("ZCN-ABC", digest)
-
-
-class ComputeNodeConnectionTests(unittest.IsolatedAsyncioTestCase):
-    async def test_new_connection_replaces_old_without_removing_new(self) -> None:
-        manager = NodeConnectionManager()
-        node_id = uuid4()
-        old = AsyncMock()
-        new = AsyncMock()
-
-        await manager.activate(node_id, old)
-        await manager.activate(node_id, new)
-
-        old.close.assert_awaited_once_with(
-            code=4001, reason="replaced by a newer connection"
-        )
-        self.assertFalse(await manager.remove(node_id, old))
-        self.assertTrue(await manager.remove(node_id, new))
-
-    async def test_admin_close_removes_and_closes_connection(self) -> None:
-        manager = NodeConnectionManager()
-        node_id = uuid4()
-        websocket = AsyncMock()
-        await manager.activate(node_id, websocket)
-        await manager.close(node_id)
-        websocket.close.assert_awaited_once_with(code=4003, reason="node disabled")
-        self.assertFalse(await manager.remove(node_id, websocket))
-
-
-class ComputeNodeMigrationTests(unittest.TestCase):
-    def test_0030_upgrade_compiles_as_postgresql_ddl(self) -> None:
-        statements: list[str] = []
-
-        def capture(sql, *multiparams, **params):
-            statements.append(str(sql.compile(dialect=postgresql.dialect())))
-
-        engine = create_mock_engine("postgresql+psycopg://", capture)
-        operations = Operations(MigrationContext.configure(engine.connect()))
-        migration = importlib.import_module(
-            "db.migrations.versions.20260812_2000_0030_compute_nodes"
-        )
-        with patch.object(migration, "op", operations):
-            migration.upgrade()
-
-        rendered = "\n".join(statements)
-        self.assertIn("compute_node_enrollments", rendered)
-        self.assertIn("compute_nodes", rendered)
-        self.assertIn("ix_compute_nodes_status", rendered)
-
-
-class ComputeNodeDeleteTests(unittest.TestCase):
-    @patch("core.compute_nodes.session_scope")
-    def test_delete_node_removes_existing_identity(self, session_scope) -> None:
-        session = session_scope.return_value.__enter__.return_value
-        node = object()
-        session.get.return_value = node
-
-        self.assertTrue(delete_node(uuid4()))
-        session.delete.assert_called_once_with(node)
-
-    @patch("core.compute_nodes.session_scope")
-    def test_delete_node_reports_missing_identity(self, session_scope) -> None:
-        session = session_scope.return_value.__enter__.return_value
-        session.get.return_value = None
-
-        self.assertFalse(delete_node(uuid4()))
-        session.delete.assert_not_called()
-
-
-if __name__ == "__main__":
-    unittest.main()

tests/test_origin_worker.py

@@ -0,0 +1,46 @@
+from __future__ import annotations
+
+import importlib.util
+import json
+import tempfile
+import unittest
+from pathlib import Path
+
+
+WORKER_PATH = (
+    Path(__file__).resolve().parents[1] / "windows-node" / "origin-worker" / "worker.py"
+)
+SPEC = importlib.util.spec_from_file_location("zcbot_origin_worker", WORKER_PATH)
+assert SPEC and SPEC.loader
+worker = importlib.util.module_from_spec(SPEC)
+SPEC.loader.exec_module(worker)
+
+
+class OriginWorkerUnitTests(unittest.TestCase):
+    def test_csv_and_json_inputs_are_read_without_origin(self) -> None:
+        with tempfile.TemporaryDirectory() as directory:
+            root = Path(directory)
+            csv_path = root / "input.csv"
+            csv_path.write_text("x,y\n1,2\n3,4\n", encoding="utf-8")
+            self.assertEqual(worker._read_rows(csv_path, None), (["x", "y"], [["1", "2"], ["3", "4"]]))
+
+            json_path = root / "input.json"
+            json_path.write_text(json.dumps([{"x": 1, "y": 2}, {"x": 3, "y": 4}]), encoding="utf-8")
+            self.assertEqual(worker._read_rows(json_path, None), (["x", "y"], [[1, 2], [3, 4]]))
+
+    def test_manifest_uses_stable_id_and_streaming_digest(self) -> None:
+        with tempfile.TemporaryDirectory() as directory:
+            path = Path(directory) / "plot-spec.json"
+            path.write_text("{}", encoding="utf-8")
+            manifest = worker._manifest(path, "application/json")
+            self.assertEqual(manifest["artifact_id"], "plot_spec")
+            self.assertEqual(manifest["sha256"], worker._file_sha256(path))
+            self.assertEqual(manifest["size_bytes"], 2)
+
+    def test_axis_title_includes_units(self) -> None:
+        self.assertEqual(worker._axis_title({"title": "Stress", "unit": "MPa"}, "Y"), "Stress (MPa)")
+        self.assertEqual(worker._axis_title(None, "Time"), "Time")
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_software_job_tools.py

@@ -0,0 +1,80 @@
+from __future__ import annotations
+
+import json
+import unittest
+from unittest.mock import patch
+from uuid import uuid4
+
+from tools.software_jobs import (
+    SoftwareCapabilityListTool,
+    SoftwareJobCancelTool,
+    SoftwareJobStatusTool,
+    SoftwareJobSubmitTool,
+)
+
+
+class SoftwareJobToolTests(unittest.TestCase):
+    def setUp(self):
+        self.user_id = uuid4()
+        self.task_id = uuid4()
+
+    def test_capability_list_reports_current_capacity(self):
+        nodes = [
+            {
+                "status": "online",
+                "capabilities": ["origin.plot@v1"],
+                "runtime": {"available_slots": 1},
+            },
+            {
+                "status": "offline",
+                "capabilities": ["origin.plot@v1"],
+                "runtime": {"available_slots": 1},
+            },
+        ]
+        with patch("tools.software_jobs.list_nodes", return_value=nodes):
+            result = json.loads(
+                SoftwareCapabilityListTool(self.user_id, self.task_id).execute()
+            )
+        self.assertEqual(result["capabilities"][0]["available_nodes"], 1)
+
+    def test_submit_injects_current_user_and_task(self):
+        created = {"job_id": str(uuid4()), "status": "queued"}
+        tool = SoftwareJobSubmitTool(self.user_id, self.task_id)
+        with patch("tools.software_jobs.create_job", return_value=(created, True)) as create:
+            result = json.loads(tool.execute("origin.plot@v1", {"input_id": str(uuid4())}))
+        self.assertTrue(result["created"])
+        self.assertEqual(create.call_args.args[:2], (self.user_id, self.task_id))
+        self.assertEqual(create.call_args.kwargs["capability"], "origin.plot@v1")
+
+    def test_status_and_cancel_reject_cross_task_job(self):
+        foreign = {"job_id": str(uuid4()), "task_id": str(uuid4())}
+        with patch("tools.software_jobs.get_job", return_value=foreign):
+            status = SoftwareJobStatusTool(self.user_id, self.task_id).execute(
+                foreign["job_id"]
+            )
+            cancel = SoftwareJobCancelTool(self.user_id, self.task_id).execute(
+                foreign["job_id"]
+            )
+        self.assertIn("not found", status)
+        self.assertIn("not found", cancel)
+
+    def test_cancel_uses_user_scoped_service(self):
+        job_id = uuid4()
+        current = {"job_id": str(job_id), "task_id": str(self.task_id)}
+        cancelled = {**current, "status": "cancelled"}
+        with (
+            patch("tools.software_jobs.get_job", return_value=current),
+            patch(
+                "tools.software_jobs.request_job_cancel",
+                return_value=(cancelled, None),
+            ) as request_cancel,
+        ):
+            result = json.loads(
+                SoftwareJobCancelTool(self.user_id, self.task_id).execute(str(job_id))
+            )
+        self.assertEqual(result["status"], "cancelled")
+        request_cancel.assert_called_once_with(self.user_id, job_id)
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_software_nodes.py

@@ -0,0 +1,379 @@
+from __future__ import annotations
+
+import importlib
+import unittest
+from pathlib import Path
+from unittest.mock import AsyncMock, patch
+from uuid import uuid4
+
+from alembic.migration import MigrationContext
+from alembic.operations import Operations
+from sqlalchemy import create_mock_engine
+from sqlalchemy.dialects import postgresql
+
+from core.software_nodes import (
+    _enrollment_digest,
+    _hash_secret,
+    _verify_secret,
+    delete_node,
+)
+from core.software_jobs import (
+    _canonical_request,
+    abandon_offer,
+    mark_node_jobs_disconnected,
+    record_job_terminal,
+    respond_to_offer,
+    update_job_state,
+    validate_output_manifest,
+    list_jobs,
+    request_job_cancel,
+)
+from web.routers.software_nodes import NodeConnectionManager, _bearer
+
+
+class SoftwareNodeSecurityTests(unittest.TestCase):
+    def test_secret_hash_is_salted_and_verifiable(self) -> None:
+        first = _hash_secret("node-secret")
+        second = _hash_secret("node-secret")
+        self.assertNotEqual(first, second)
+        self.assertNotIn("node-secret", first)
+        self.assertTrue(_verify_secret("node-secret", first))
+        self.assertFalse(_verify_secret("wrong", first))
+
+    def test_bearer_parser_rejects_query_style_or_missing_token(self) -> None:
+        self.assertEqual(_bearer("Bearer abc"), "abc")
+        with self.assertRaisesRegex(Exception, "missing node bearer token"):
+            _bearer(None)
+
+    def test_enrollment_digest_does_not_store_plaintext(self) -> None:
+        digest = _enrollment_digest("ZCN-ABC")
+        self.assertEqual(len(digest), 64)
+        self.assertNotIn("ZCN-ABC", digest)
+
+    def test_websocket_auth_rejection_uses_explicit_application_close_code(self) -> None:
+        source = (
+            Path(__file__).resolve().parents[1] / "web" / "routers" / "software_nodes.py"
+        ).read_text(encoding="utf-8")
+        rejection = source.split("except (ValueError, SoftwareNodeError):", 1)[1].split(
+            "await node_connections.activate", 1
+        )[0]
+        self.assertLess(
+            rejection.index("await websocket.accept()"), rejection.index("await websocket.close")
+        )
+        self.assertIn('code=4003, reason="invalid node credentials"', rejection)
+
+
+class SoftwareNodeConnectionTests(unittest.IsolatedAsyncioTestCase):
+    async def test_new_connection_replaces_old_without_removing_new(self) -> None:
+        manager = NodeConnectionManager()
+        node_id = uuid4()
+        old = AsyncMock()
+        new = AsyncMock()
+
+        await manager.activate(node_id, old)
+        await manager.activate(node_id, new)
+
+        old.close.assert_awaited_once_with(
+            code=4001, reason="replaced by a newer connection"
+        )
+        self.assertFalse(await manager.remove(node_id, old))
+        self.assertTrue(await manager.remove(node_id, new))
+
+    async def test_admin_close_removes_and_closes_connection(self) -> None:
+        manager = NodeConnectionManager()
+        node_id = uuid4()
+        websocket = AsyncMock()
+        await manager.activate(node_id, websocket)
+        await manager.close(node_id)
+        websocket.close.assert_awaited_once_with(code=4003, reason="node disabled")
+        self.assertFalse(await manager.remove(node_id, websocket))
+
+
+class SoftwareNodeMigrationTests(unittest.TestCase):
+    def test_0030_upgrade_compiles_as_postgresql_ddl(self) -> None:
+        statements: list[str] = []
+
+        def capture(sql, *multiparams, **params):
+            statements.append(str(sql.compile(dialect=postgresql.dialect())))
+
+        engine = create_mock_engine("postgresql+psycopg://", capture)
+        operations = Operations(MigrationContext.configure(engine.connect()))
+        migration = importlib.import_module(
+            "db.migrations.versions.20260812_2000_0030_compute_nodes"
+        )
+        with patch.object(migration, "op", operations):
+            migration.upgrade()
+
+        rendered = "\n".join(statements)
+        self.assertIn("compute_node_enrollments", rendered)
+        self.assertIn("compute_nodes", rendered)
+        self.assertIn("ix_compute_nodes_status", rendered)
+
+    def test_0032_upgrade_compiles_as_postgresql_ddl(self) -> None:
+        statements: list[str] = []
+
+        def capture(sql, *multiparams, **params):
+            statements.append(str(sql.compile(dialect=postgresql.dialect())))
+
+        engine = create_mock_engine("postgresql+psycopg://", capture)
+        operations = Operations(MigrationContext.configure(engine.connect()))
+        migration = importlib.import_module(
+            "db.migrations.versions.20260813_1600_0032_software_jobs"
+        )
+        with patch.object(migration, "op", operations):
+            migration.upgrade()
+
+        rendered = "\n".join(statements)
+        self.assertIn("ALTER TABLE compute_node_enrollments RENAME TO software_node_enrollments", rendered)
+        self.assertIn("ALTER TABLE compute_nodes RENAME TO software_nodes", rendered)
+        self.assertIn("software_jobs", rendered)
+        self.assertIn("uq_software_jobs_user_idempotency", rendered)
+        self.assertIn("ix_software_jobs_status_created", rendered)
+
+
+class SoftwareJobProtocolTests(unittest.TestCase):
+    @patch("core.software_jobs.session_scope")
+    def test_queued_job_cancels_without_node_message(self, session_scope) -> None:
+        session = session_scope.return_value.__enter__.return_value
+        job = type("Job", (), {})()
+        job.job_id = uuid4(); job.user_id = uuid4(); job.task_id = uuid4()
+        job.capability = "origin.plot@v1"; job.request_digest = "a" * 64
+        job.node_id = None; job.lease_id = None; job.status = "queued"; job.stage = ""
+        job.progress = 0; job.metrics = {}; job.error = {}; job.artifact_manifest = []
+        job.created_at = None; job.started_at = None; job.terminal_at = None
+        session.execute.return_value.scalar_one_or_none.return_value = job
+        result, node_message = request_job_cancel(job.user_id, job.job_id)
+        self.assertEqual(result["status"], "cancelled")
+        self.assertIsNone(node_message)
+        self.assertEqual(job.error["code"], "USER_CANCELLED")
+
+    @patch("core.software_jobs.session_scope")
+    def test_running_job_persists_cancel_before_sending(self, session_scope) -> None:
+        session = session_scope.return_value.__enter__.return_value
+        job = type("Job", (), {})()
+        job.job_id = uuid4(); job.user_id = uuid4(); job.task_id = uuid4()
+        job.capability = "origin.plot@v1"; job.request_digest = "b" * 64
+        job.node_id = uuid4(); job.lease_id = uuid4(); job.status = "running"
+        job.stage = "origin_running"; job.progress = 10; job.metrics = {}; job.error = {}
+        job.artifact_manifest = []; job.created_at = None; job.started_at = None; job.terminal_at = None
+        session.execute.return_value.scalar_one_or_none.return_value = job
+        result, node_message = request_job_cancel(job.user_id, job.job_id)
+        self.assertEqual(result["status"], "cancelling")
+        self.assertEqual(node_message["node_id"], job.node_id)
+        self.assertEqual(node_message["payload"]["lease_id"], str(job.lease_id))
+
+    @patch("core.software_jobs.session_scope")
+    def test_job_list_is_enriched_for_job_center(self, session_scope) -> None:
+        session = session_scope.return_value.__enter__.return_value
+        job = type("Job", (), {})()
+        job.job_id = uuid4(); job.task_id = uuid4(); job.capability = "origin.plot@v1"
+        job.request_digest = "c" * 64; job.node_id = uuid4(); job.status = "running"
+        job.stage = "origin_running"; job.progress = 20; job.metrics = {}; job.error = {}
+        job.artifact_manifest = []; job.input_manifest = {"filename": "input.xlsx"}
+        job.request = {"plot": {"title": "Test"}, "output": {"formats": ["png"]}}
+        job.created_at = None; job.started_at = None; job.terminal_at = None
+        session.execute.return_value.all.return_value = [(job, "材料仿真", "LAB-01")]
+        results = list_jobs(uuid4(), limit=10)
+        self.assertEqual(results[0]["task_name"], "材料仿真")
+        self.assertEqual(results[0]["node_name"], "LAB-01")
+        self.assertEqual(results[0]["request_summary"]["display_name"], "Origin 科研绘图")
+
+    def test_origin_request_is_canonical_and_rejects_extra_fields(self) -> None:
+        request = {
+            "schema_version": 1,
+            "input": {"input_id": str(uuid4()), "sheet": "Sheet1"},
+            "plot": {"type": "line", "x": "x", "y": ["y"]},
+            "output": {"formats": ["png", "opju"]},
+        }
+        normalized, digest = _canonical_request(request)
+        self.assertEqual(normalized, request)
+        self.assertEqual(len(digest), 64)
+        with self.assertRaisesRegex(Exception, "invalid origin plot request fields"):
+            _canonical_request({**request, "script": "anything"})
+        with self.assertRaisesRegex(Exception, "unsupported origin plot fields"):
+            _canonical_request({**request, "plot": {**request["plot"], "script": "anything"}})
+        with self.assertRaisesRegex(Exception, "input.input_id must be an artifact UUID"):
+            _canonical_request({**request, "input": {"input_id": "C:\\data.csv"}})
+
+    def test_origin_request_rejects_unimplemented_plot_semantics(self) -> None:
+        request = {
+            "schema_version": 1,
+            "input": {"input_id": str(uuid4())},
+            "plot": {"type": "scatter", "x": "time", "y": ["a", "b"]},
+            "output": {"formats": ["png"], "dpi": 300},
+        }
+        for plot, message in (
+            ({**request["plot"], "template": "custom"}, "unsupported origin plot template"),
+            ({**request["plot"], "x_axis": {"scale": "log10"}}, "invalid x_axis"),
+            ({**request["plot"], "legend": {"enabled": False}}, "invalid plot.legend"),
+            ({**request["plot"], "y": ["a", "a"]}, "plot.y must contain"),
+        ):
+            with self.subTest(message=message), self.assertRaisesRegex(Exception, message):
+                _canonical_request({**request, "plot": plot})
+        with self.assertRaisesRegex(Exception, "video recording is not supported"):
+            _canonical_request(
+                {**request, "output": {"formats": ["png"], "record_video": True}}
+            )
+
+    def test_output_manifest_matches_exact_requested_formats(self) -> None:
+        request = {"output": {"formats": ["opju", "png"]}}
+        manifest = [
+            {"artifact_id": "project", "filename": "project.opju", "media_type": "application/x-origin-project", "size_bytes": 10, "sha256": "a" * 64},
+            {"artifact_id": "figure_png", "filename": "figure.png", "media_type": "image/png", "size_bytes": 20, "sha256": "b" * 64},
+            {"artifact_id": "plot_spec", "filename": "plot-spec.json", "media_type": "application/json", "size_bytes": 30, "sha256": "c" * 64},
+            {"artifact_id": "provenance", "filename": "provenance.json", "media_type": "application/json", "size_bytes": 40, "sha256": "d" * 64},
+        ]
+        self.assertEqual(validate_output_manifest(request, manifest), manifest)
+        with self.assertRaisesRegex(Exception, "incomplete"):
+            validate_output_manifest(request, manifest[:-1])
+        with self.assertRaisesRegex(Exception, "metadata"):
+            validate_output_manifest(
+                request,
+                [{**manifest[0], "filename": "anything.opju"}, *manifest[1:]],
+            )
+
+    @patch("core.software_jobs.session_scope")
+    def test_stale_offer_cannot_be_accepted_by_another_node(self, session_scope) -> None:
+        session = session_scope.return_value.__enter__.return_value
+        job = type("Job", (), {})()
+        job.node_id = uuid4()
+        job.lease_id = uuid4()
+        job.status = "offered"
+        session.execute.return_value.scalar_one_or_none.return_value = job
+        with self.assertRaisesRegex(Exception, "stale or does not belong"):
+            respond_to_offer(
+                uuid4(),
+                accepted=True,
+                payload={"job_id": str(uuid4()), "lease_id": str(job.lease_id)},
+            )
+
+    @patch("core.software_jobs.session_scope")
+    def test_failed_delivery_only_abandons_matching_offer(self, session_scope) -> None:
+        session = session_scope.return_value.__enter__.return_value
+        node_id = uuid4()
+        lease_id = uuid4()
+        job = type("Job", (), {})()
+        job.node_id = node_id
+        job.lease_id = lease_id
+        job.status = "offered"
+        session.execute.return_value.scalar_one_or_none.return_value = job
+        abandon_offer(
+            node_id,
+            {"job_id": str(uuid4()), "lease_id": str(lease_id)},
+        )
+        self.assertEqual(job.status, "queued")
+        self.assertIsNone(job.node_id)
+
+    def test_dispatcher_excludes_nodes_with_active_jobs(self) -> None:
+        source = (
+            Path(__file__).resolve().parents[1] / "core" / "software_jobs.py"
+        ).read_text(encoding="utf-8")
+        self.assertIn('{"offered", "dispatched", "running"}', source)
+        self.assertIn("item.node_id not in busy_node_ids", source)
+
+    def test_input_download_rechecks_file_digest(self) -> None:
+        source = (
+            Path(__file__).resolve().parents[1]
+            / "web" / "routers" / "software_nodes.py"
+        ).read_text(encoding="utf-8")
+        self.assertIn("digest = sha256()", source)
+        self.assertIn('digest.hexdigest() != item["sha256"]', source)
+
+    @patch("core.software_jobs.session_scope")
+    def test_job_state_restores_disconnected_job(self, session_scope) -> None:
+        session = session_scope.return_value.__enter__.return_value
+        node_id = uuid4()
+        lease_id = uuid4()
+        digest = "a" * 64
+        job = type("Job", (), {})()
+        job.node_id = node_id
+        job.lease_id = lease_id
+        job.request_digest = digest
+        job.status = "disconnected"
+        job.started_at = None
+        session.execute.return_value.scalar_one_or_none.return_value = job
+        update_job_state(node_id, {
+            "job_id": str(uuid4()),
+            "lease_id": str(lease_id),
+            "request_digest": digest,
+            "stage": "waiting_input",
+            "progress": 0,
+            "metrics": {},
+        })
+        self.assertEqual(job.status, "dispatched")
+        self.assertEqual(job.stage, "waiting_input")
+
+    @patch("core.software_jobs.session_scope")
+    def test_ready_to_run_is_not_reported_as_running(self, session_scope) -> None:
+        session = session_scope.return_value.__enter__.return_value
+        node_id = uuid4()
+        lease_id = uuid4()
+        digest = "c" * 64
+        job = type("Job", (), {})()
+        job.node_id = node_id
+        job.lease_id = lease_id
+        job.request_digest = digest
+        job.status = "dispatched"
+        job.started_at = None
+        session.execute.return_value.scalar_one_or_none.return_value = job
+        update_job_state(node_id, {
+            "job_id": str(uuid4()), "lease_id": str(lease_id),
+            "request_digest": digest, "stage": "ready_to_run",
+            "progress": 5, "metrics": {"input_bytes": 10},
+        })
+        self.assertEqual(job.status, "dispatched")
+
+    @patch("core.software_jobs.session_scope")
+    def test_terminal_replay_is_idempotent(self, session_scope) -> None:
+        session = session_scope.return_value.__enter__.return_value
+        node_id = uuid4()
+        lease_id = uuid4()
+        digest = "b" * 64
+        job = type("Job", (), {})()
+        job.node_id = node_id
+        job.lease_id = lease_id
+        job.request_digest = digest
+        job.status = "failed"
+        session.execute.return_value.scalar_one_or_none.return_value = job
+        record_job_terminal(node_id, {
+            "job_id": str(uuid4()),
+            "lease_id": str(lease_id),
+            "request_digest": digest,
+            "status": "failed",
+            "error": {"code": "TEST"},
+            "artifact_manifest": [],
+        })
+        self.assertEqual(job.status, "failed")
+
+    @patch("core.software_jobs.session_scope")
+    def test_disconnect_does_not_requeue_active_jobs(self, session_scope) -> None:
+        session = session_scope.return_value.__enter__.return_value
+        first = type("Job", (), {"status": "running"})()
+        second = type("Job", (), {"status": "dispatched"})()
+        session.execute.return_value.scalars.return_value = [first, second]
+        mark_node_jobs_disconnected(uuid4())
+        self.assertEqual(first.status, "disconnected")
+        self.assertEqual(second.status, "disconnected")
+
+
+class SoftwareNodeDeleteTests(unittest.TestCase):
+    @patch("core.software_nodes.session_scope")
+    def test_delete_node_removes_existing_identity(self, session_scope) -> None:
+        session = session_scope.return_value.__enter__.return_value
+        node = object()
+        session.get.return_value = node
+
+        self.assertTrue(delete_node(uuid4()))
+        session.delete.assert_called_once_with(node)
+
+    @patch("core.software_nodes.session_scope")
+    def test_delete_node_reports_missing_identity(self, session_scope) -> None:
+        session = session_scope.return_value.__enter__.return_value
+        session.get.return_value = None
+
+        self.assertFalse(delete_node(uuid4()))
+        session.delete.assert_not_called()
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_software_output_publish.py

@@ -0,0 +1,65 @@
+from __future__ import annotations
+
+import hashlib
+import tempfile
+import unittest
+from pathlib import Path
+from unittest.mock import patch
+from uuid import uuid4
+
+from web.routers.software_nodes import _publish_software_job_outputs
+
+
+class SoftwareOutputPublishTests(unittest.TestCase):
+    def test_complete_set_moves_atomically_and_can_be_replayed(self) -> None:
+        with tempfile.TemporaryDirectory() as directory:
+            root = Path(directory)
+            job_id = uuid4()
+            working_dir = root / "research"
+            staging = root / ".zcbot_software_job_staging" / str(job_id)
+            staging.mkdir(parents=True)
+            working_dir.mkdir()
+            content = b"origin-result"
+            (staging / "figure.png").write_bytes(content)
+            manifest = [{
+                "artifact_id": "figure_png",
+                "filename": "figure.png",
+                "media_type": "image/png",
+                "size_bytes": len(content),
+                "sha256": hashlib.sha256(content).hexdigest(),
+            }]
+            context = {
+                "user_id": uuid4(),
+                "task_id": uuid4(),
+                "working_dir": "research",
+            }
+
+            def register(**kwargs):
+                return tuple({
+                    "version": 2,
+                    "scope": "working_dir",
+                    "path": ref["path"],
+                    "label": ref["label"],
+                    "artifact_id": str(uuid4()),
+                } for ref in kwargs["refs"])
+
+            with (
+                patch("web.routers.software_nodes.load_user_root", return_value=root),
+                patch(
+                    "web.routers.software_nodes.register_published_artifacts",
+                    side_effect=register,
+                ),
+            ):
+                first = _publish_software_job_outputs(job_id, context, manifest)
+                second = _publish_software_job_outputs(job_id, context, manifest)
+
+            published = working_dir / "origin" / str(job_id) / "figure.png"
+            self.assertEqual(published.read_bytes(), content)
+            self.assertFalse(staging.exists())
+            self.assertEqual(first[0]["source_artifact_id"], "figure_png")
+            self.assertEqual(first[0]["path"], f"origin/{job_id}/figure.png")
+            self.assertEqual(second[0]["source_artifact_id"], "figure_png")
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_static_vendor.py

@@ -26,19 +26,29 @@ def _frontend_source() -> str:
 
 
 class StaticVendorTests(unittest.TestCase):
+    def test_dev_console_has_software_job_center(self) -> None:
+        html = DEV_HTML.read_text(encoding="utf-8")
+        source = (JS_DIR / "software_jobs.js").read_text(encoding="utf-8")
+        self.assertIn('id="software-job-center"', html)
+        self.assertIn('id="software-job-toggle"', html)
+        self.assertIn('"/v1/software-jobs?limit=50"', source)
+        self.assertIn("/cancel`", source)
+        self.assertIn("分析结果", source)
+
     def test_admin_can_create_windows_node_enrollment_code(self) -> None:
         html = ADMIN_HTML.read_text(encoding="utf-8")
         admin_js = (JS_DIR / "admin.js").read_text(encoding="utf-8")
 
         self.assertIn('id="node-enrollment-modal" class="modal"', html)
         self.assertIn("生成 Windows Node 注册码", html)
-        self.assertIn('"/v1/admin/compute-node-enrollments"', admin_js)
+        self.assertIn('"/v1/admin/software-node-enrollments"', admin_js)
         self.assertIn('capabilities: ["origin.plot@v1"]', admin_js)
+        self.assertIn('origin.health === "ready"', admin_js)
         self.assertIn("ttl_seconds: 600", admin_js)
         self.assertIn("navigator.clipboard.writeText(value)", admin_js)
-        self.assertIn('apiGet("/v1/admin/compute-nodes")', admin_js)
-        self.assertIn('apiSend("PATCH", `/v1/admin/compute-nodes/${node.node_id}`', admin_js)
-        self.assertIn('apiSend("DELETE", `/v1/admin/compute-nodes/${node.node_id}`', admin_js)
+        self.assertIn('apiGet("/v1/admin/software-nodes")', admin_js)
+        self.assertIn('apiSend("PATCH", `/v1/admin/software-nodes/${node.node_id}`', admin_js)
+        self.assertIn('apiSend("DELETE", `/v1/admin/software-nodes/${node.node_id}`', admin_js)
         self.assertIn("最近心跳", admin_js)
         self.assertIn("重新启用", admin_js)
         self.assertIn("永久删除", admin_js)

tests/test_web_routes_nodb.py

@@ -110,8 +110,10 @@ class AuthGateTests(unittest.TestCase):
         ("POST", "/v1/tasks"),
         ("POST", "/v1/asr/transcribe"),
         ("GET", "/v1/admin/overview"),
-        ("GET", "/v1/admin/compute-nodes"),
-        ("DELETE", "/v1/admin/compute-nodes/00000000-0000-0000-0000-000000000000"),
+        ("GET", "/v1/admin/software-nodes"),
+        ("GET", "/v1/software-jobs"),
+        ("POST", "/v1/software-jobs/00000000-0000-0000-0000-000000000000/cancel"),
+        ("DELETE", "/v1/admin/software-nodes/00000000-0000-0000-0000-000000000000"),
         ("GET", "/v1/admin/tool-wire-health"),
         ("GET", "/v1/admin/external-system-definitions"),
         ("GET", "/v1/admin/external-system-users"),

tests/test_windows_node_source.py

@@ -20,8 +20,8 @@ class WindowsNodeSourceTests(unittest.TestCase):
     def test_node_protocol_and_secret_storage_markers_are_present(self) -> None:
         source = "\n".join(path.read_text(encoding="utf-8") for path in PROJECT.glob("*.cs"))
         for marker in (
-            "v1/compute/nodes/enroll",
-            "v1/compute/nodes/connect",
+            "v1/software-nodes/enroll",
+            "v1/software-nodes/connect",
             'SetRequestHeader("Authorization"',
             'SetRequestHeader("X-Node-Id"',
             "DataProtectionScope.LocalMachine",
@@ -35,7 +35,11 @@ class WindowsNodeSourceTests(unittest.TestCase):
             self.assertIn(marker, source)
 
     def test_node_does_not_expose_arbitrary_execution_primitives(self) -> None:
-        source = "\n".join(path.read_text(encoding="utf-8") for path in PROJECT.glob("*.cs"))
+        source = "\n".join(
+            path.read_text(encoding="utf-8")
+            for path in PROJECT.glob("*.cs")
+            if path.name != "OriginWorkerRunner.cs"
+        )
         for forbidden in ("Process.Start", "cmd.exe", "powershell.exe", "LabTalk"):
             self.assertNotIn(forbidden, source)
 
@@ -56,7 +60,7 @@ class WindowsNodeSourceTests(unittest.TestCase):
 
     def test_configuration_window_is_resizable_and_dpi_safe(self) -> None:
         form = (PROJECT / "ConfigurationForm.cs").read_text(encoding="utf-8")
-        self.assertIn("ClientSize = new Size(880, 720)", form)
+        self.assertIn("ClientSize = new Size(840, 680)", form)
         self.assertIn("FormBorderStyle.Sizable", form)
         self.assertIn("AutoScaleMode.Dpi", form)
         self.assertIn("AutoScroll = true", form)
@@ -65,12 +69,17 @@ class WindowsNodeSourceTests(unittest.TestCase):
         self.assertIn("成功注册一次后立即失效", form)
         self.assertIn("CreateCard", form)
         self.assertIn("注册并连接", form)
-        self.assertIn("ContentWidth = 800", form)
-        self.assertIn("节点能力", form)
+        self.assertIn("ContentWidth = 760", form)
         self.assertIn("Origin 绘图", form)
-        self.assertIn("当前 MVP 内置声明该协议，不需要手工配置", form)
+        self.assertIn('CreateCapabilityRow("Origin 绘图", "origin.plot@v1")', form)
+        self.assertIn('CreateButton("立即重连", 112, primary: true)', form)
+        self.assertIn("ReconnectRequested?.Invoke()", form)
         self.assertIn("registrationCard.Visible = !registered", form)
-        self.assertIn("registeredActionsCard.Visible = registered", form)
+        self.assertIn("reconnect.Visible = registered", form)
+        self.assertIn("resetIdentity.Visible = registered", form)
+
+        tray = (PROJECT / "TrayApplicationContext.cs").read_text(encoding="utf-8")
+        self.assertIn("form.ReconnectRequested += RestartConnection", tray)
 
     def test_startup_task_is_login_scoped_and_runs_the_fixed_node_executable(self) -> None:
         script = (ROOT / "install-startup.ps1").read_text(encoding="utf-8")
@@ -80,6 +89,109 @@ class WindowsNodeSourceTests(unittest.TestCase):
         self.assertIn("-RunLevel Limited", script)
         self.assertNotIn("-RunLevel Highest", script)
 
+    def test_auth_rejection_is_distinct_from_http_websocket_handshake_failure(self) -> None:
+        connection = (PROJECT / "NodeConnectionLoop.cs").read_text(encoding="utf-8")
+        self.assertIn('socket.HttpStatusCode is HttpStatusCode.Unauthorized', connection)
+        self.assertIn('or HttpStatusCode.Forbidden', connection)
+        self.assertIn("WebSocket 握手被拒绝，请检查服务端或反向代理", connection)
+        self.assertIn("catch (NodeEndpointException exception)", connection)
+        self.assertIn("throw new NodeEndpointException", connection)
+        self.assertIn("(int?)result.CloseStatus == 4003", connection)
+        self.assertIn("节点身份已被服务端拒绝", connection)
+        self.assertNotIn("Node credentials were rejected", connection)
+
+    def test_origin_runtime_probe_is_read_only_and_reported(self) -> None:
+        probe = (PROJECT / "OriginRuntimeProbe.cs").read_text(encoding="utf-8")
+        connection = (PROJECT / "NodeConnectionLoop.cs").read_text(encoding="utf-8")
+        self.assertIn('AutomationProgId = @"Origin.ApplicationSI\\CLSID"', probe)
+        self.assertIn("RegistryHive.LocalMachine", probe)
+        self.assertIn("RegistryHive.CurrentUser", probe)
+        self.assertIn('new("OriginPro", version, "0.2.0", health, detail)', probe)
+        self.assertIn(
+            'available_slots = origin.Health == "ready" && !jobInbox.HasPendingJobs ? 1 : 0',
+            connection,
+        )
+        self.assertNotIn("CreateInstance", probe)
+        self.assertNotIn("Process.Start", probe)
+        for marker in (
+            "software_version = origin.SoftwareVersion",
+            "adapter_version = origin.AdapterVersion",
+            "health = origin.Health",
+            "detail = origin.Detail",
+        ):
+            self.assertIn(marker, connection)
+
+    def test_job_offer_is_persisted_before_acceptance(self) -> None:
+        inbox = (PROJECT / "JobInboxStore.cs").read_text(encoding="utf-8")
+        connection = (PROJECT / "NodeConnectionLoop.cs").read_text(encoding="utf-8")
+        self.assertIn('capabilityValue.GetString() != "origin.plot@v1"', inbox)
+        self.assertIn("PlotTypes.Contains", inbox)
+        self.assertIn("OutputFormats.Contains", inbox)
+        self.assertIn("FileOptions.WriteThrough", inbox)
+        self.assertIn("stream.Flush(flushToDisk: true)", inbox)
+        new_record = inbox.split("var record =", 1)[1].split("private static JsonElement?", 1)[0]
+        self.assertLess(
+            new_record.index("AtomicWrite(requestPath, record"),
+            new_record.index("JobOfferResult.Accept"),
+        )
+        self.assertIn('offerResult.Accepted ? "job_accept" : "job_reject"', connection)
+        self.assertIn("sendLock.WaitAsync", connection)
+        self.assertIn("!jobInbox.HasPendingJobs ? 1 : 0", connection)
+        self.assertIn("ReportRecoverableJobsAsync", connection)
+        self.assertIn("ConcurrentDictionary<Guid, Task> jobPipelines", connection)
+        self.assertIn("StartJobPipeline(socket, acceptedJob)", connection)
+        self.assertIn("inputDownloader.DownloadAsync(job, CancellationToken.None)", connection)
+        self.assertIn('stage = "uploading_outputs"', connection)
+        self.assertIn("&& !job.UploadComplete", connection)
+        self.assertIn("StartJobPipeline(socket, job)", connection)
+        self.assertIn('stage = "waiting_input"', connection)
+        self.assertIn('Path.Combine(jobDirectory, "terminal.json")', inbox)
+        self.assertIn("AtomicWrite(requestPath, updated, overwrite: true)", inbox)
+
+        downloader = (PROJECT / "JobInputDownloader.cs").read_text(encoding="utf-8")
+        self.assertIn('new AuthenticationHeaderValue("Bearer", config.NodeToken)', downloader)
+        self.assertIn('DefaultRequestHeaders.Add("X-Node-Id"', downloader)
+        self.assertIn("HttpCompletionOption.ResponseHeadersRead", downloader)
+        self.assertIn("IncrementalHash.CreateHash", downloader)
+        self.assertIn("total > expectedSize", downloader)
+        self.assertIn("File.Move(temporaryPath, destination, overwrite: false)", downloader)
+        self.assertNotIn("Process.Start", downloader)
+
+    def test_origin_worker_launch_is_fixed_and_terminal_driven(self) -> None:
+        runner = (PROJECT / "OriginWorkerRunner.cs").read_text(encoding="utf-8")
+        connection = (PROJECT / "NodeConnectionLoop.cs").read_text(encoding="utf-8")
+        project = (PROJECT / "Zcbot.WindowsNode.csproj").read_text(encoding="utf-8")
+        worker = (ROOT / "origin-worker" / "worker.py").read_text(encoding="utf-8")
+        self.assertIn('Environment.GetEnvironmentVariable("ZCBOT_ORIGIN_PYTHON")', runner)
+        self.assertIn('Path.Combine(paths.RootDirectory, "runtimes", "origin", "python.exe")', runner)
+        self.assertIn("UseShellExecute = false", runner)
+        self.assertIn("startInfo.ArgumentList.Add(workerScript)", runner)
+        self.assertIn("startInfo.ArgumentList.Add(jobDirectory)", runner)
+        self.assertIn('Path.Combine(jobDirectory, "terminal.json")', runner)
+        self.assertIn('"NODE_RESTARTED_DURING_JOB"', runner)
+        self.assertIn("CancellationTokenSource.CreateLinkedTokenSource", runner)
+        self.assertIn("process.Kill(entireProcessTree: true)", runner)
+        self.assertIn('type.GetString() == "job_cancel"', connection)
+        self.assertIn('"cancelled", "USER_CANCELLED"', connection)
+        self.assertIn("origin-worker\\worker.py", project)
+        self.assertIn("if op.oext:", worker)
+        self.assertIn("op.exit()", worker)
+        self.assertIn("op.new_graph", worker)
+        self.assertIn("layer.add_plot", worker)
+        self.assertIn("op.save", worker)
+        self.assertIn("graph.save_fig", worker)
+        self.assertIn('_atomic_json(job_dir / "terminal.json"', worker)
+        for forbidden in ("subprocess", "eval(", "exec(", "os.system"):
+            self.assertNotIn(forbidden, worker)
+
+        uploader = (PROJECT / "JobOutputUploader.cs").read_text(encoding="utf-8")
+        self.assertIn('new AuthenticationHeaderValue("Bearer", config.NodeToken)', uploader)
+        self.assertIn('DefaultRequestHeaders.Add("X-Node-Id"', uploader)
+        self.assertIn('DefaultRequestHeaders.Add("X-Lease-Id"', uploader)
+        self.assertIn("SHA256.HashDataAsync", uploader)
+        self.assertIn("upload-complete.json", connection + uploader)
+        self.assertNotIn("Process.Start", uploader)
+
 
 if __name__ == "__main__":
     unittest.main()

tools/software_jobs.py

@@ -0,0 +1,122 @@
+"""Agent 可调用的专业软件任务工具。"""
+from __future__ import annotations
+
+import json
+from uuid import UUID, uuid4
+
+from core.software_jobs import (
+    SoftwareJobError,
+    create_job,
+    get_job,
+    list_jobs,
+    request_job_cancel,
+)
+from core.software_nodes import SUPPORTED_CAPABILITIES, list_nodes
+from .base import Tool
+
+
+class _SoftwareJobTool(Tool):
+    def __init__(self, user_id: UUID, task_id: UUID, **kwargs) -> None:
+        super().__init__(**kwargs)
+        self.user_id = user_id
+        self.task_id = task_id
+
+
+class SoftwareCapabilityListTool(_SoftwareJobTool):
+    name = "software_capability_list"
+    description = "List professional software capabilities available through managed Windows nodes."
+    parameters = {"type": "object", "properties": {}, "additionalProperties": False}
+
+    def execute(self) -> str:
+        nodes = list_nodes()
+        items = [{
+            "capability": item,
+            "display_name": "Origin 科研绘图" if item == "origin.plot@v1" else item,
+            "available_nodes": sum(
+                1
+                for node in nodes
+                if node["status"] == "online"
+                and item in (node.get("capabilities") or [])
+                and (node.get("runtime") or {}).get("available_slots", 0) > 0
+            ),
+        } for item in sorted(SUPPORTED_CAPABILITIES)]
+        return json.dumps({"capabilities": items}, ensure_ascii=False)
+
+
+class SoftwareJobSubmitTool(_SoftwareJobTool):
+    name = "software_job_submit"
+    description = (
+        "Submit a persistent professional software job for the current task. "
+        "Return immediately with job_id; do not poll continuously or wait for completion."
+    )
+    parameters = {
+        "type": "object",
+        "properties": {
+            "capability": {"type": "string", "enum": sorted(SUPPORTED_CAPABILITIES)},
+            "request": {"type": "object"},
+            "idempotency_key": {
+                "type": "string",
+                "description": "Stable unique key for this exact submission; omit to generate one.",
+            },
+        },
+        "required": ["capability", "request"],
+        "additionalProperties": False,
+    }
+
+    def execute(self, capability: str, request: dict, idempotency_key: str = "") -> str:
+        try:
+            job, created = create_job(
+                self.user_id,
+                self.task_id,
+                idempotency_key=idempotency_key.strip() or str(uuid4()),
+                capability=capability,
+                request=request,
+            )
+            return json.dumps({**job, "created": created}, ensure_ascii=False)
+        except SoftwareJobError as exc:
+            return f"[Error] {exc}"
+
+
+class SoftwareJobStatusTool(_SoftwareJobTool):
+    name = "software_job_status"
+    description = "Check one software job, or list recent jobs in the current task when job_id is omitted."
+    parameters = {
+        "type": "object",
+        "properties": {"job_id": {"type": "string"}},
+        "additionalProperties": False,
+    }
+
+    def execute(self, job_id: str = "") -> str:
+        try:
+            if job_id.strip():
+                item = get_job(self.user_id, UUID(job_id.strip()))
+                if item is None or item["task_id"] != str(self.task_id):
+                    return "[Error] software job not found"
+                return json.dumps(item, ensure_ascii=False)
+            return json.dumps(
+                {"results": list_jobs(self.user_id, task_id=self.task_id, limit=20)},
+                ensure_ascii=False,
+            )
+        except ValueError:
+            return "[Error] invalid job_id"
+
+
+class SoftwareJobCancelTool(_SoftwareJobTool):
+    name = "software_job_cancel"
+    description = "Request cancellation of a software job in the current task after the user asks to stop it."
+    parameters = {
+        "type": "object",
+        "properties": {"job_id": {"type": "string"}},
+        "required": ["job_id"],
+        "additionalProperties": False,
+    }
+
+    def execute(self, job_id: str) -> str:
+        try:
+            item = get_job(self.user_id, UUID(job_id.strip()))
+            if item is None or item["task_id"] != str(self.task_id):
+                return "[Error] software job not found"
+            job, _ = request_job_cancel(self.user_id, UUID(job_id.strip()))
+            return json.dumps(job, ensure_ascii=False)
+        except (ValueError, SoftwareJobError) as exc:
+            return f"[Error] {exc}"

web/app.py

@@ -49,7 +49,7 @@ from .background import (
 from .broker import broker
 from .routers.asr import register_asr_routes
 from .routers.authroutes import register_auth_routes
-from .routers.compute_nodes import register_compute_node_routes
+from .routers.software_nodes import register_software_node_routes
 from .routers.external_systems import register_external_system_routes
 from .routers.files import register_file_routes
 from .routers.kb import register_kb_routes
@@ -206,7 +206,9 @@ def create_app() -> FastAPI:
     register_asr_routes(app, require_user=require_user, auth_cfg=auth_cfg)
     register_task_routes(app, require_user=require_user)
     register_message_routes(app, require_user=require_user)
-    register_compute_node_routes(app, require_admin=require_admin)
+    register_software_node_routes(
+        app, require_user=require_user, require_admin=require_admin
+    )
 
     # ───────────── 管理后台(admin-only)─────────────
     register_admin_routes(app, require_admin)

web/routers/compute_nodes.py

@@ -1,158 +0,0 @@
-"""Windows Node MVP 的注册、管理与长连接端点。"""
-
-from __future__ import annotations
-
-import asyncio
-from uuid import UUID
-
-from fastapi import Depends, HTTPException, WebSocket, WebSocketDisconnect, status
-
-from core.compute_nodes import (
-    ComputeNodeError,
-    authenticate_node,
-    create_enrollment,
-    delete_node,
-    enroll_node,
-    list_nodes,
-    mark_node_offline,
-    set_node_disabled,
-    update_node_runtime,
-)
-from web.schemas import (
-    ComputeEnrollmentCreateRequest,
-    ComputeNodeDisableRequest,
-    ComputeNodeEnrollRequest,
-)
-
-
-class NodeConnectionManager:
-    def __init__(self) -> None:
-        self._connections: dict[UUID, WebSocket] = {}
-        self._lock = asyncio.Lock()
-
-    async def activate(self, node_id: UUID, websocket: WebSocket) -> None:
-        async with self._lock:
-            old = self._connections.get(node_id)
-            self._connections[node_id] = websocket
-        if old is not None and old is not websocket:
-            await old.close(code=4001, reason="replaced by a newer connection")
-
-    async def remove(self, node_id: UUID, websocket: WebSocket) -> bool:
-        async with self._lock:
-            if self._connections.get(node_id) is websocket:
-                self._connections.pop(node_id, None)
-                return True
-        return False
-
-    async def close(self, node_id: UUID) -> None:
-        async with self._lock:
-            websocket = self._connections.pop(node_id, None)
-        if websocket is not None:
-            await websocket.close(code=4003, reason="node disabled")
-
-
-node_connections = NodeConnectionManager()
-
-
-def _bearer(authorization: str | None) -> str:
-    scheme, _, token = (authorization or "").partition(" ")
-    if scheme.lower() != "bearer" or not token:
-        raise ComputeNodeError("missing node bearer token")
-    return token
-
-
-def register_compute_node_routes(app, *, require_admin) -> None:
-    @app.post(
-        "/v1/compute/nodes/enroll",
-        tags=["compute-nodes"],
-        status_code=status.HTTP_201_CREATED,
-    )
-    def node_enroll(body: ComputeNodeEnrollRequest):
-        try:
-            return enroll_node(**body.model_dump())
-        except ComputeNodeError as exc:
-            raise HTTPException(400, str(exc)) from exc
-
-    @app.websocket("/v1/compute/nodes/connect")
-    async def node_connect(websocket: WebSocket):
-        try:
-            node_id = UUID(websocket.headers.get("x-node-id", ""))
-            token = _bearer(websocket.headers.get("authorization"))
-            identity = await asyncio.to_thread(authenticate_node, node_id, token)
-        except (ValueError, ComputeNodeError):
-            await websocket.close(code=1008, reason="invalid node credentials")
-            return
-        await websocket.accept()
-        await node_connections.activate(node_id, websocket)
-        try:
-            await websocket.send_json({"type": "connected", "heartbeat_seconds": 15})
-            while True:
-                message = await websocket.receive_json()
-                message_type = message.get("type")
-                payload = message.get("payload") or {}
-                if message_type not in {"hello", "heartbeat"} or not isinstance(
-                    payload, dict
-                ):
-                    await websocket.send_json(
-                        {"type": "error", "code": "unsupported_message"}
-                    )
-                    continue
-                if payload.get("install_id") and payload["install_id"] != str(
-                    identity["install_id"]
-                ):
-                    await websocket.close(code=1008, reason="install identity mismatch")
-                    return
-                await asyncio.to_thread(
-                    update_node_runtime,
-                    node_id,
-                    status="online",
-                    runtime=payload,
-                )
-                await websocket.send_json(
-                    {"type": "ack", "message_id": message.get("message_id")}
-                )
-        except (ComputeNodeError, WebSocketDisconnect, RuntimeError, ValueError):
-            pass
-        finally:
-            if await node_connections.remove(node_id, websocket):
-                await asyncio.to_thread(mark_node_offline, node_id)
-
-    @app.post("/v1/admin/compute-node-enrollments", tags=["admin"])
-    def admin_create_compute_enrollment(
-        body: ComputeEnrollmentCreateRequest,
-        user_id: UUID = Depends(require_admin),  # noqa: B008
-    ):
-        try:
-            return create_enrollment(user_id, **body.model_dump())
-        except ComputeNodeError as exc:
-            raise HTTPException(400, str(exc)) from exc
-
-    @app.get("/v1/admin/compute-nodes", tags=["admin"])
-    def admin_compute_nodes(user_id: UUID = Depends(require_admin)):  # noqa: B008
-        return {"results": list_nodes()}
-
-    @app.patch("/v1/admin/compute-nodes/{node_id}", tags=["admin"])
-    async def admin_disable_compute_node(
-        node_id: UUID,
-        body: ComputeNodeDisableRequest,
-        user_id: UUID = Depends(require_admin),  # noqa: B008
-    ):
-        if not await asyncio.to_thread(set_node_disabled, node_id, body.disabled):
-            raise HTTPException(404, "compute node not found")
-        if body.disabled:
-            await node_connections.close(node_id)
-        return {
-            "node_id": str(node_id),
-            "status": "disabled" if body.disabled else "offline",
-        }
-
-    @app.delete("/v1/admin/compute-nodes/{node_id}", tags=["admin"])
-    async def admin_delete_compute_node(
-        node_id: UUID,
-        user_id: UUID = Depends(require_admin),  # noqa: B008
-    ):
-        # 先撤掉在线连接，避免删除后的旧 socket 继续上报运行态。
-        await node_connections.close(node_id)
-        if not await asyncio.to_thread(delete_node, node_id):
-            raise HTTPException(404, "compute node not found")
-        return {"node_id": str(node_id), "status": "deleted"}

web/routers/software_nodes.py

@@ -0,0 +1,576 @@
+"""Windows Node MVP 的注册、管理与长连接端点。"""
+
+from __future__ import annotations
+
+import asyncio
+import os
+from hashlib import sha256
+from pathlib import Path
+from uuid import UUID
+
+from fastapi import Depends, Header, HTTPException, Request, WebSocket, WebSocketDisconnect, status
+from fastapi.responses import FileResponse
+
+from core.software_nodes import (
+    SoftwareNodeError,
+    authenticate_node,
+    create_enrollment,
+    delete_node,
+    enroll_node,
+    list_nodes,
+    mark_node_offline,
+    set_node_disabled,
+    update_node_runtime,
+)
+from core.software_jobs import (
+    MAX_OUTPUT_ARTIFACT_BYTES,
+    MAX_OUTPUT_TOTAL_BYTES,
+    OUTPUT_ARTIFACTS,
+    abandon_offer,
+    create_job,
+    get_job,
+    get_job_input,
+    get_job_output_context,
+    mark_node_jobs_disconnected,
+    offer_next_job,
+    pending_node_cancellations,
+    record_job_terminal,
+    list_jobs,
+    request_job_cancel,
+    respond_to_offer,
+    update_job_state,
+    validate_output_manifest,
+    SoftwareJobError,
+)
+from core.artifact_lifecycle import register_published_artifacts
+from web.schemas import (
+    SoftwareEnrollmentCreateRequest,
+    SoftwareJobCreateRequest,
+    SoftwareNodeDisableRequest,
+    SoftwareNodeEnrollRequest,
+)
+from web.userfiles import load_user_root, safe_join
+
+
+class NodeConnectionManager:
+    def __init__(self) -> None:
+        self._connections: dict[UUID, WebSocket] = {}
+        self._send_locks: dict[UUID, asyncio.Lock] = {}
+        self._lock = asyncio.Lock()
+
+    async def activate(self, node_id: UUID, websocket: WebSocket) -> None:
+        async with self._lock:
+            old = self._connections.get(node_id)
+            self._connections[node_id] = websocket
+            self._send_locks.setdefault(node_id, asyncio.Lock())
+        if old is not None and old is not websocket:
+            await old.close(code=4001, reason="replaced by a newer connection")
+
+    async def remove(self, node_id: UUID, websocket: WebSocket) -> bool:
+        async with self._lock:
+            if self._connections.get(node_id) is websocket:
+                self._connections.pop(node_id, None)
+                self._send_locks.pop(node_id, None)
+                return True
+        return False
+
+    async def close(self, node_id: UUID) -> None:
+        async with self._lock:
+            websocket = self._connections.pop(node_id, None)
+            self._send_locks.pop(node_id, None)
+        if websocket is not None:
+            await websocket.close(code=4003, reason="node disabled")
+
+    async def node_ids(self) -> set[UUID]:
+        async with self._lock:
+            return set(self._connections)
+
+    async def send(self, node_id: UUID, message: dict) -> bool:
+        async with self._lock:
+            websocket = self._connections.get(node_id)
+            send_lock = self._send_locks.get(node_id)
+        if websocket is None or send_lock is None:
+            return False
+        async with send_lock:
+            await websocket.send_json(message)
+        return True
+
+    async def send_on(self, node_id: UUID, websocket: WebSocket, message: dict) -> bool:
+        async with self._lock:
+            current = self._connections.get(node_id)
+            send_lock = self._send_locks.get(node_id)
+        if current is not websocket or send_lock is None:
+            return False
+        async with send_lock:
+            await websocket.send_json(message)
+        return True
+
+
+node_connections = NodeConnectionManager()
+
+
+def _bearer(authorization: str | None) -> str:
+    scheme, _, token = (authorization or "").partition(" ")
+    if scheme.lower() != "bearer" or not token:
+        raise SoftwareNodeError("missing node bearer token")
+    return token
+
+
+def _authenticate_output_request(
+    job_id: UUID,
+    authorization: str | None,
+    x_node_id: str,
+    x_lease_id: str,
+    x_request_digest: str,
+) -> tuple[UUID, UUID, dict]:
+    try:
+        node_id = UUID(x_node_id)
+        lease_id = UUID(x_lease_id)
+        authenticate_node(node_id, _bearer(authorization))
+    except (ValueError, SoftwareNodeError) as exc:
+        raise HTTPException(401, "invalid node credentials or job identity") from exc
+    context = get_job_output_context(node_id, job_id, lease_id, x_request_digest)
+    if context is None:
+        raise HTTPException(404, "software job output target not found")
+    return node_id, lease_id, context
+
+
+def _hash_file(path: Path) -> str:
+    digest = sha256()
+    with path.open("rb") as handle:
+        for chunk in iter(lambda: handle.read(1024 * 1024), b""):
+            digest.update(chunk)
+    return digest.hexdigest()
+
+
+def _reject_symlink_path(root: Path, target: Path) -> None:
+    root = root.resolve()
+    current = root
+    for part in target.relative_to(root).parts:
+        current = current / part
+        if current.is_symlink():
+            raise HTTPException(409, "software job output path contains a symbolic link")
+
+
+def _publish_software_job_outputs(job_id: UUID, context: dict, manifest: list[dict]) -> list[dict]:
+    root = load_user_root(context["user_id"])
+    working_dir = safe_join(root, context["working_dir"])
+    staging = safe_join(root, f".zcbot_software_job_staging/{job_id}")
+    relative_output = Path("origin") / str(job_id)
+    destination = safe_join(working_dir, relative_output.as_posix())
+    source = staging if staging.is_dir() else destination
+    _reject_symlink_path(root, source)
+    _reject_symlink_path(root, destination)
+    for item in manifest:
+        path = source / item["filename"]
+        if (
+            not path.is_file()
+            or path.stat().st_size != item["size_bytes"]
+            or _hash_file(path) != item["sha256"]
+        ):
+            raise SoftwareJobError(f"uploaded artifact is missing or invalid: {item['artifact_id']}")
+    if source == staging:
+        destination.parent.mkdir(parents=True, exist_ok=True)
+        if destination.exists():
+            raise SoftwareJobError("software job output destination already exists unexpectedly")
+        os.replace(staging, destination)
+        try:
+            staging.parent.rmdir()
+        except OSError:
+            pass
+    refs = tuple({
+        "path": (relative_output / item["filename"]).as_posix(),
+        "label": item["filename"],
+        "media_type": item["media_type"],
+    } for item in manifest)
+    published_refs = register_published_artifacts(
+        user_id=context["user_id"],
+        task_id=context["task_id"],
+        user_root=root,
+        working_dir=working_dir,
+        refs=refs,
+    )
+    refs_by_path = {item["path"]: item for item in published_refs}
+    return [
+        {
+            **item,
+            "source_artifact_id": item["artifact_id"],
+            "artifact_id": refs_by_path[(relative_output / item["filename"]).as_posix()]["artifact_id"],
+            "path": (relative_output / item["filename"]).as_posix(),
+        }
+        for item in manifest
+    ]
+
+
+def register_software_node_routes(app, *, require_user, require_admin) -> None:
+    @app.post(
+        "/v1/software-nodes/enroll",
+        tags=["software-nodes"],
+        status_code=status.HTTP_201_CREATED,
+    )
+    def node_enroll(body: SoftwareNodeEnrollRequest):
+        try:
+            return enroll_node(**body.model_dump())
+        except SoftwareNodeError as exc:
+            raise HTTPException(400, str(exc)) from exc
+
+    @app.get("/v1/software-jobs/{job_id}/input", tags=["software-nodes"])
+    def download_software_job_input(
+        job_id: UUID,
+        authorization: str | None = Header(default=None),
+        x_node_id: str = Header(default=""),
+    ):
+        try:
+            node_id = UUID(x_node_id)
+            authenticate_node(node_id, _bearer(authorization))
+        except (ValueError, SoftwareNodeError) as exc:
+            raise HTTPException(401, "invalid node credentials") from exc
+        item = get_job_input(node_id, job_id)
+        if item is None:
+            raise HTTPException(404, "software job input not found")
+        target = safe_join(load_user_root(item["user_id"]), item["current_path"])
+        if not target.is_file():
+            raise HTTPException(404, "software job input file not found")
+        stat = target.stat()
+        if stat.st_size != item["size_bytes"]:
+            raise HTTPException(409, "software job input changed after submission")
+        digest = sha256()
+        with target.open("rb") as handle:
+            for chunk in iter(lambda: handle.read(1024 * 1024), b""):
+                digest.update(chunk)
+        if digest.hexdigest() != item["sha256"]:
+            raise HTTPException(409, "software job input changed after submission")
+        return FileResponse(
+            path=str(target),
+            filename=item["filename"],
+            media_type="application/octet-stream",
+            headers={
+                "Cache-Control": "no-store",
+                "X-Content-SHA256": item["sha256"],
+            },
+        )
+
+    @app.put(
+        "/v1/software-jobs/{job_id}/outputs/{artifact_id}",
+        tags=["software-nodes"],
+        status_code=status.HTTP_204_NO_CONTENT,
+    )
+    async def upload_software_job_output(
+        job_id: UUID,
+        artifact_id: str,
+        request: Request,
+        authorization: str | None = Header(default=None),
+        x_node_id: str = Header(default=""),
+        x_lease_id: str = Header(default=""),
+        x_request_digest: str = Header(default=""),
+        x_content_sha256: str = Header(default=""),
+        x_content_length: int = Header(default=-1),
+    ):
+        _, _, context = await asyncio.to_thread(
+            _authenticate_output_request,
+            job_id, authorization, x_node_id, x_lease_id, x_request_digest,
+        )
+        metadata = OUTPUT_ARTIFACTS.get(artifact_id)
+        if metadata is None:
+            raise HTTPException(400, "unsupported output artifact identity")
+        filename, _, output_format = metadata
+        requested_formats = set(context["request"].get("output", {}).get("formats") or [])
+        if output_format is not None and output_format not in requested_formats:
+            raise HTTPException(400, "output artifact was not requested")
+        if not 1 <= x_content_length <= MAX_OUTPUT_ARTIFACT_BYTES:
+            raise HTTPException(400, "output artifact size is invalid")
+        if len(x_content_sha256) != 64 or any(c not in "0123456789abcdef" for c in x_content_sha256):
+            raise HTTPException(400, "output artifact digest is invalid")
+        root = load_user_root(context["user_id"])
+        published = safe_join(
+            safe_join(root, context["working_dir"]),
+            f"origin/{job_id}/{filename}",
+        )
+        if published.is_file():
+            if published.stat().st_size == x_content_length and _hash_file(published) == x_content_sha256:
+                return None
+            raise HTTPException(409, "published output conflicts with uploaded artifact")
+        staging = safe_join(root, f".zcbot_software_job_staging/{job_id}")
+        _reject_symlink_path(root, staging)
+        staging.mkdir(parents=True, exist_ok=True)
+        destination = staging / filename
+        if destination.is_file():
+            if destination.stat().st_size == x_content_length and _hash_file(destination) == x_content_sha256:
+                return None
+            raise HTTPException(409, "uploaded output conflicts with existing staging file")
+        staged_total = sum(
+            item.stat().st_size for item in staging.iterdir() if item.is_file()
+        )
+        if staged_total + x_content_length > MAX_OUTPUT_TOTAL_BYTES:
+            raise HTTPException(413, "software job outputs exceed the total size limit")
+        temporary = destination.with_name(destination.name + ".tmp-" + os.urandom(8).hex())
+        digest = sha256()
+        total = 0
+        try:
+            with temporary.open("xb") as handle:
+                async for chunk in request.stream():
+                    total += len(chunk)
+                    if total > x_content_length or total > MAX_OUTPUT_ARTIFACT_BYTES:
+                        raise HTTPException(413, "output artifact exceeded declared size")
+                    digest.update(chunk)
+                    handle.write(chunk)
+                handle.flush()
+                os.fsync(handle.fileno())
+            if total != x_content_length or digest.hexdigest() != x_content_sha256:
+                raise HTTPException(400, "output artifact did not match declared metadata")
+            os.replace(temporary, destination)
+        finally:
+            temporary.unlink(missing_ok=True)
+        return None
+
+    @app.post("/v1/software-jobs/{job_id}/outputs/complete", tags=["software-nodes"])
+    async def complete_software_job_outputs(
+        job_id: UUID,
+        request: Request,
+        authorization: str | None = Header(default=None),
+        x_node_id: str = Header(default=""),
+        x_lease_id: str = Header(default=""),
+        x_request_digest: str = Header(default=""),
+    ):
+        node_id, lease_id, context = await asyncio.to_thread(
+            _authenticate_output_request,
+            job_id, authorization, x_node_id, x_lease_id, x_request_digest,
+        )
+        body = await request.json()
+        if not isinstance(body, dict):
+            raise HTTPException(400, "output completion body must be an object")
+        try:
+            manifest = validate_output_manifest(context["request"], body.get("artifact_manifest"))
+            published = await asyncio.to_thread(
+                _publish_software_job_outputs, job_id, context, manifest
+            )
+            terminal = {
+                "job_id": str(job_id),
+                "lease_id": str(lease_id),
+                "request_digest": x_request_digest,
+                "status": "succeeded",
+                "error": {},
+                "artifact_manifest": published,
+            }
+            await asyncio.to_thread(record_job_terminal, node_id, terminal)
+        except (SoftwareJobError, KeyError, TypeError) as exc:
+            raise HTTPException(409, str(exc)) from exc
+        return {"status": "succeeded", "artifact_manifest": published}
+
+    @app.websocket("/v1/software-nodes/connect")
+    async def node_connect(websocket: WebSocket):
+        try:
+            node_id = UUID(websocket.headers.get("x-node-id", ""))
+            token = _bearer(websocket.headers.get("authorization"))
+            identity = await asyncio.to_thread(authenticate_node, node_id, token)
+        except (ValueError, SoftwareNodeError):
+            # 握手前 close 会被 ASGI 统一表现为 HTTP 403，客户端无法区分
+            # “凭据无效”和“代理/路由没有正确转发 WebSocket”。先升级再用
+            # 应用关闭码给已持有 Node ID/Token 的节点返回明确诊断。
+            await websocket.accept()
+            await websocket.close(code=4003, reason="invalid node credentials")
+            return
+        await websocket.accept()
+        await node_connections.activate(node_id, websocket)
+        try:
+            await node_connections.send_on(
+                node_id,
+                websocket,
+                {"type": "connected", "heartbeat_seconds": 15},
+            )
+            for cancel in await asyncio.to_thread(pending_node_cancellations, node_id):
+                await node_connections.send_on(
+                    node_id, websocket, {"type": "job_cancel", "payload": cancel}
+                )
+            while True:
+                message = await websocket.receive_json()
+                message_type = message.get("type")
+                payload = message.get("payload") or {}
+                if not isinstance(payload, dict):
+                    await node_connections.send_on(node_id, websocket,
+                        {"type": "error", "code": "unsupported_message"}
+                    )
+                    continue
+                if message_type in {"job_accept", "job_reject"}:
+                    await asyncio.to_thread(
+                        respond_to_offer,
+                        node_id,
+                        accepted=message_type == "job_accept",
+                        payload=payload,
+                    )
+                    await node_connections.send_on(node_id, websocket,
+                        {"type": "ack", "message_id": message.get("message_id")}
+                    )
+                    continue
+                if message_type in {"job_state", "job_terminal"}:
+                    if message_type == "job_terminal" and payload.get("status") == "succeeded":
+                        await node_connections.send_on(node_id, websocket,
+                            {
+                                "type": "error",
+                                "code": "outputs_not_published",
+                                "message_id": message.get("message_id"),
+                            }
+                        )
+                        continue
+                    handler = (
+                        update_job_state
+                        if message_type == "job_state"
+                        else record_job_terminal
+                    )
+                    await asyncio.to_thread(handler, node_id, payload)
+                    await node_connections.send_on(node_id, websocket,
+                        {"type": "ack", "message_id": message.get("message_id")}
+                    )
+                    continue
+                if message_type not in {"hello", "heartbeat"}:
+                    await node_connections.send_on(node_id, websocket,
+                        {"type": "error", "code": "unsupported_message"}
+                    )
+                    continue
+                if payload.get("install_id") and payload["install_id"] != str(
+                    identity["install_id"]
+                ):
+                    await websocket.close(code=1008, reason="install identity mismatch")
+                    return
+                await asyncio.to_thread(
+                    update_node_runtime,
+                    node_id,
+                    status="online",
+                    runtime=payload,
+                )
+                await node_connections.send_on(node_id, websocket,
+                    {"type": "ack", "message_id": message.get("message_id")}
+                )
+                for cancel in await asyncio.to_thread(pending_node_cancellations, node_id):
+                    await node_connections.send_on(
+                        node_id, websocket, {"type": "job_cancel", "payload": cancel}
+                    )
+                offer = await asyncio.to_thread(
+                    offer_next_job, await node_connections.node_ids()
+                )
+                if offer is not None:
+                    delivered = await node_connections.send(
+                        offer["node_id"],
+                        {"type": "job_offer", "payload": offer["payload"]},
+                    )
+                    if not delivered:
+                        await asyncio.to_thread(
+                            abandon_offer, offer["node_id"], offer["payload"]
+                        )
+        except (
+            SoftwareJobError,
+            SoftwareNodeError,
+            WebSocketDisconnect,
+            RuntimeError,
+            ValueError,
+        ):
+            pass
+        finally:
+            if await node_connections.remove(node_id, websocket):
+                await asyncio.to_thread(mark_node_offline, node_id)
+                await asyncio.to_thread(mark_node_jobs_disconnected, node_id)
+
+    @app.post("/v1/tasks/{task_id}/software-jobs", tags=["software-jobs"])
+    async def submit_software_job(
+        task_id: UUID,
+        body: SoftwareJobCreateRequest,
+        user_id: UUID = Depends(require_user),  # noqa: B008
+    ):
+        try:
+            job, created = await asyncio.to_thread(
+                create_job, user_id, task_id, **body.model_dump()
+            )
+            offer = await asyncio.to_thread(
+                offer_next_job, await node_connections.node_ids()
+            )
+            if offer is not None:
+                delivered = await node_connections.send(
+                    offer["node_id"], {"type": "job_offer", "payload": offer["payload"]}
+                )
+                if not delivered:
+                    await asyncio.to_thread(
+                        abandon_offer, offer["node_id"], offer["payload"]
+                    )
+            return {**job, "created": created}
+        except SoftwareJobError as exc:
+            detail = str(exc)
+            raise HTTPException(404 if detail == "task not found" else 400, detail) from exc
+
+    @app.get("/v1/software-jobs", tags=["software-jobs"])
+    def read_software_jobs(
+        task_id: UUID | None = None,
+        active_only: bool = False,
+        limit: int = 50,
+        user_id: UUID = Depends(require_user),  # noqa: B008
+    ):
+        return {
+            "results": list_jobs(
+                user_id, task_id=task_id, active_only=active_only, limit=limit
+            )
+        }
+
+    @app.post("/v1/software-jobs/{job_id}/cancel", tags=["software-jobs"])
+    async def cancel_software_job(
+        job_id: UUID,
+        user_id: UUID = Depends(require_user),  # noqa: B008
+    ):
+        try:
+            job, message = await asyncio.to_thread(request_job_cancel, user_id, job_id)
+        except SoftwareJobError as exc:
+            raise HTTPException(404, str(exc)) from exc
+        if message is not None:
+            await node_connections.send(
+                message["node_id"],
+                {"type": "job_cancel", "payload": message["payload"]},
+            )
+        return job
+
+    @app.get("/v1/software-jobs/{job_id}", tags=["software-jobs"])
+    def read_software_job(
+        job_id: UUID,
+        user_id: UUID = Depends(require_user),  # noqa: B008
+    ):
+        job = get_job(user_id, job_id)
+        if job is None:
+            raise HTTPException(404, "software job not found")
+        return job
+
+    @app.post("/v1/admin/software-node-enrollments", tags=["admin"])
+    def admin_create_software_enrollment(
+        body: SoftwareEnrollmentCreateRequest,
+        user_id: UUID = Depends(require_admin),  # noqa: B008
+    ):
+        try:
+            return create_enrollment(user_id, **body.model_dump())
+        except SoftwareNodeError as exc:
+            raise HTTPException(400, str(exc)) from exc
+
+    @app.get("/v1/admin/software-nodes", tags=["admin"])
+    def admin_software_nodes(user_id: UUID = Depends(require_admin)):  # noqa: B008
+        return {"results": list_nodes()}
+
+    @app.patch("/v1/admin/software-nodes/{node_id}", tags=["admin"])
+    async def admin_disable_software_node(
+        node_id: UUID,
+        body: SoftwareNodeDisableRequest,
+        user_id: UUID = Depends(require_admin),  # noqa: B008
+    ):
+        if not await asyncio.to_thread(set_node_disabled, node_id, body.disabled):
+            raise HTTPException(404, "software node not found")
+        if body.disabled:
+            await node_connections.close(node_id)
+        return {
+            "node_id": str(node_id),
+            "status": "disabled" if body.disabled else "offline",
+        }
+
+    @app.delete("/v1/admin/software-nodes/{node_id}", tags=["admin"])
+    async def admin_delete_software_node(
+        node_id: UUID,
+        user_id: UUID = Depends(require_admin),  # noqa: B008
+    ):
+        # 先撤掉在线连接，避免删除后的旧 socket 继续上报运行态。
+        await node_connections.close(node_id)
+        if not await asyncio.to_thread(delete_node, node_id):
+            raise HTTPException(404, "software node not found")
+        return {"node_id": str(node_id), "status": "deleted"}

web/schemas.py

@@ -116,13 +116,13 @@ class ExternalSystemCredentialsRequest(BaseModel):
     credentials: dict[str, str] = Field(default_factory=dict)
 
 
-class ComputeEnrollmentCreateRequest(BaseModel):
+class SoftwareEnrollmentCreateRequest(BaseModel):
     expected_name: str = ""
     capabilities: list[str] = Field(default_factory=lambda: ["origin.plot@v1"])
     ttl_seconds: int = 600
 
 
-class ComputeNodeEnrollRequest(BaseModel):
+class SoftwareNodeEnrollRequest(BaseModel):
     enrollment_code: str
     node_name: str
     install_id: UUID
@@ -131,5 +131,11 @@ class ComputeNodeEnrollRequest(BaseModel):
     capabilities: list[str]
 
 
-class ComputeNodeDisableRequest(BaseModel):
+class SoftwareNodeDisableRequest(BaseModel):
     disabled: bool = True
+
+
+class SoftwareJobCreateRequest(BaseModel):
+    idempotency_key: str
+    capability: str = "origin.plot@v1"
+    request: dict = Field(default_factory=dict)

web/static/dev.html

@@ -1069,6 +1069,34 @@
     overflow: hidden; text-overflow: ellipsis; white-space: nowrap; margin-top: 2px;
   }
   .proc-toast.fail { border-color: rgba(192,57,43,0.5); }
+  #software-job-center {
+    position: fixed; right: 16px; bottom: 16px; z-index: 115; display: none;
+    width: min(380px, calc(100vw - 24px)); font-size: 13px;
+  }
+  #software-job-center.show { display: block; }
+  #software-job-toggle {
+    margin-left: auto; display: flex; align-items: center; gap: 8px;
+    border-radius: 999px; padding: 8px 13px; background: #fff;
+    border: 1px solid var(--border); box-shadow: 0 4px 16px rgba(0,0,0,.16); cursor: pointer;
+  }
+  #software-job-center.expanded #software-job-toggle { border-radius: 0 0 10px 10px; }
+  #software-job-panel {
+    max-height: min(520px, calc(100vh - 110px)); overflow: auto; background: #fff;
+    border: 1px solid var(--border); border-bottom: 0; border-radius: 10px 10px 0 0;
+    box-shadow: 0 -4px 20px rgba(0,0,0,.16); padding: 8px;
+  }
+  .sj-card { padding: 10px; border-bottom: 1px solid var(--border); }
+  .sj-card:last-child { border-bottom: 0; }
+  .sj-title { display: flex; justify-content: space-between; gap: 10px; }
+  .sj-title span { color: var(--muted); white-space: nowrap; }
+  .sj-card.failed .sj-title span { color: var(--danger); }
+  .sj-card.disconnected .sj-title span, .sj-card.cancelling .sj-title span { color: #a66514; }
+  .sj-sub, .sj-meta { margin-top: 4px; color: var(--muted); font-size: 11px; overflow-wrap: anywhere; }
+  .sj-progress { height: 4px; margin-top: 7px; border-radius: 4px; background: var(--panel-muted); overflow: hidden; }
+  .sj-progress i { display: block; height: 100%; background: var(--accent); transition: width .25s ease; }
+  .sj-actions { display: flex; justify-content: flex-end; gap: 6px; margin-top: 8px; }
+  .sj-empty { padding: 22px; text-align: center; color: var(--muted); }
+  #software-job-center.show ~ #proc-toasts { bottom: 62px; }
   /* media tool 摘要 banner(model / size / cost / elapsed,折叠态也可见) */
   .tool-banner {
     display: inline-flex; flex-wrap: wrap; gap: 6px;
@@ -2174,6 +2202,13 @@
   </div>
 </div>
 
+<div id="software-job-center" aria-live="polite">
+  <div id="software-job-panel" hidden></div>
+  <button id="software-job-toggle" type="button" title="展开专业软件任务">
+    <span>专业软件任务</span><strong id="software-job-count">0 条</strong>
+  </button>
+</div>
+
 <script type="module" src="js/main.js"></script>
 </body>
 </html>

web/static/js/admin.js

@@ -55,7 +55,7 @@ let externalDefinitions = [];
 let externalUsers = [];
 let externalDefinitionsLoaded = false;
 let externalEditingId = "";
-let computeNodes = [];
+let softwareNodes = [];
 
 // ───── 格式化 ─────
 function fmtCNY(n) {
@@ -167,12 +167,16 @@ function nodeStatusHTML(status) {
 }
 
 function renderWindowsNodes() {
-  const rows = computeNodes.map(node => {
+  const rows = softwareNodes.map(node => {
     const runtime = node.runtime || {};
+    const origin = runtime.origin || {};
+    const originState = origin.health === "ready" ? "Origin 可用" : "Origin 不可用";
+    const originVersion = origin.software_version ? ` ${origin.software_version}` : "";
     const runtimeParts = [
       node.os_version || "",
       runtime.desktop_session === true ? "桌面会话" : "",
       runtime.available_slots != null ? `可用槽位 ${runtime.available_slots}` : "",
+      runtime.origin ? `${originState}${originVersion}` : "",
     ].filter(Boolean);
     const lastSeen = node.last_seen_at
       ? `<span title="${escapeHtml(fmtTime(node.last_seen_at))}">${escapeHtml(fmtTimeAgo(node.last_seen_at))}</span>`
@@ -193,22 +197,22 @@ function renderWindowsNodes() {
   }).join("") || `<tr><td colspan="7" class="empty">尚无已注册的 Windows Node</td></tr>`;
 
   $("s-windows-node").innerHTML = `<div class="card"><div class="card-head">`
-    + `<div><h2>Windows Node（${computeNodes.length}）</h2><div class="node-help">查看节点状态；禁用会立即断开节点并拒绝后续连接。</div></div>`
+    + `<div><h2>Windows Node（${softwareNodes.length}）</h2><div class="node-help">查看节点状态；禁用会立即断开节点并拒绝后续连接。</div></div>`
     + `<button id="node-enrollment-open" class="primary" type="button">生成 Windows Node 注册码</button>`
     + `</div><div class="scroll-x"><table><thead><tr><th>节点</th><th>状态</th><th>运行环境</th>`
     + `<th>版本</th><th>能力</th><th>最近心跳</th><th>操作</th></tr></thead><tbody>${rows}</tbody></table></div></div>`;
   $("node-enrollment-open").onclick = openNodeEnrollmentModal;
   $("s-windows-node").querySelectorAll("[data-node-toggle]").forEach(button => {
-    button.onclick = () => toggleComputeNode(button);
+    button.onclick = () => toggleSoftwareNode(button);
   });
   $("s-windows-node").querySelectorAll("[data-node-delete]").forEach(button => {
-    button.onclick = () => deleteComputeNode(button);
+    button.onclick = () => deleteSoftwareNode(button);
   });
 }
 
-async function toggleComputeNode(button) {
+async function toggleSoftwareNode(button) {
   const row = button.closest("tr[data-node-id]");
-  const node = computeNodes.find(item => item.node_id === row?.dataset.nodeId);
+  const node = softwareNodes.find(item => item.node_id === row?.dataset.nodeId);
   if (!node) return;
   const disabling = button.dataset.nodeToggle === "disable";
   const confirmed = await dialogConfirm({
@@ -222,20 +226,20 @@ async function toggleComputeNode(button) {
   if (!confirmed) return;
   button.disabled = true;
   try {
-    await apiSend("PATCH", `/v1/admin/compute-nodes/${node.node_id}`, {
+    await apiSend("PATCH", `/v1/admin/software-nodes/${node.node_id}`, {
       disabled: disabling,
     });
     message(disabling ? "节点已禁用" : "节点已重新启用，请在节点电脑上立即重连", "success", 5000);
-    await loadComputeNodes();
+    await loadSoftwareNodes();
   } catch (err) {
     if (err.code !== "auth") message("更新节点失败：" + (err.message || String(err)), "error", 5000);
     button.disabled = false;
   }
 }
 
-async function deleteComputeNode(button) {
+async function deleteSoftwareNode(button) {
   const row = button.closest("tr[data-node-id]");
-  const node = computeNodes.find(item => item.node_id === row?.dataset.nodeId);
+  const node = softwareNodes.find(item => item.node_id === row?.dataset.nodeId);
   if (!node) return;
   const confirmed = await dialogConfirm({
     title: "删除 Windows Node",
@@ -246,9 +250,9 @@ async function deleteComputeNode(button) {
   if (!confirmed) return;
   button.disabled = true;
   try {
-    await apiSend("DELETE", `/v1/admin/compute-nodes/${node.node_id}`, {});
+    await apiSend("DELETE", `/v1/admin/software-nodes/${node.node_id}`, {});
     message("节点已删除，本机需重新注册后才能使用", "success", 5000);
-    await loadComputeNodes();
+    await loadSoftwareNodes();
   } catch (err) {
     if (err.code !== "auth") message("删除节点失败：" + (err.message || String(err)), "error", 5000);
     button.disabled = false;
@@ -306,7 +310,7 @@ async function createNodeEnrollment(e) {
   submit.disabled = true;
   submit.textContent = "生成中…";
   try {
-    const result = await apiSend("POST", "/v1/admin/compute-node-enrollments", {
+    const result = await apiSend("POST", "/v1/admin/software-node-enrollments", {
       expected_name: $("node-expected-name").value.trim(),
       capabilities: ["origin.plot@v1"],
       ttl_seconds: 600,
@@ -1031,10 +1035,10 @@ async function loadExternalDefinitions(force = false) {
   } catch (e) { /* overview 统一处理鉴权 */ }
 }
 
-async function loadComputeNodes() {
+async function loadSoftwareNodes() {
   try {
-    const result = await apiGet("/v1/admin/compute-nodes");
-    computeNodes = result.results || [];
+    const result = await apiGet("/v1/admin/software-nodes");
+    softwareNodes = result.results || [];
     renderWindowsNodes();
   } catch (e) { /* overview 统一处理鉴权 */ }
 }
@@ -1048,7 +1052,7 @@ async function refresh() {
     loadModels();
     loadUserUsage(userPage);
     loadStorage(storagePage);
-    loadComputeNodes();
+    loadSoftwareNodes();
     loadExternalDefinitions();
     loadToolFailures();
   } catch (e) {

web/static/js/main.js

@@ -19,6 +19,7 @@ import { loadFolderSuggestions } from "./newtask.js";
 import { embedInit } from "./embed.js";
 import { loadTaskList, loadModels, loadChannelCards, selectTask, showNewConversationDraft } from "./chat.js";
 import { refreshProcs } from "./procs.js";
+import { initSoftwareJobs } from "./software_jobs.js";
 
 // ───── enter app ─────
 export function enterApp(initialTaskId = "") {
@@ -40,6 +41,7 @@ export function enterApp(initialTaskId = "") {
   loadStorage(); // 顶栏存储用量(后台扫描快照,非实时)
   loadRole();    // 拉 /v1/me,admin 才显「管理」入口(/static/admin.html)
   refreshProcs(); // 后台进程(bg proc):上个会话遗留的 running proc 恢复轮询/展示
+  initSoftwareJobs();
 }
 
 // 顶栏用户名:默认显 name(兜底 user_name → email → uid8),title 悬浮给完整身份。

web/static/js/software_jobs.js

@@ -0,0 +1,144 @@
+// 专业软件任务中心：用户级轮询、跨对话状态、取消与终态通知。
+import { api } from "./api.js";
+import { state } from "./state.js";
+import { $ } from "./dom.js";
+import { escapeHtml } from "./format.js";
+import { selectTask } from "./chat.js";
+import { dialogConfirm, message } from "./dialog.js";
+
+const ACTIVE = new Set(["queued", "offered", "dispatched", "running", "disconnected", "cancelling"]);
+const TERMINAL = new Set(["succeeded", "failed", "cancelled"]);
+const POLL_ACTIVE_MS = 4000;
+const POLL_IDLE_MS = 30000;
+let timer = null;
+let known = new Map();
+let jobs = [];
+let expanded = false;
+let initialized = false;
+
+const statusLabel = {
+  queued: "等待计算节点", offered: "正在分配节点", dispatched: "节点已接收",
+  running: "正在执行", disconnected: "节点连接中断", cancelling: "正在停止",
+  succeeded: "已完成", failed: "失败", cancelled: "已取消",
+};
+const stageLabel = {
+  accepted: "节点已接收", waiting_input: "正在下载输入文件", ready_to_run: "准备软件环境",
+  origin_running: "Origin 正在生成图表", uploading_outputs: "正在上传结果",
+  cancel_requested: "停止请求已发送", terminal: "任务已结束",
+};
+
+export function initSoftwareJobs() {
+  if (initialized || !$("software-job-center")) return;
+  initialized = true;
+  $("software-job-toggle").onclick = () => { expanded = !expanded; render(); };
+  document.addEventListener("click", (event) => {
+    const center = $("software-job-center");
+    if (expanded && center && !center.contains(event.target)) { expanded = false; render(); }
+  });
+  refreshSoftwareJobs();
+}
+
+export async function refreshSoftwareJobs() {
+  if (!state.token) return;
+  try {
+    const data = await api("GET", "/v1/software-jobs?limit=50");
+    const next = data.results || [];
+    next.forEach((job) => {
+      const previous = known.get(job.job_id);
+      if (previous && ACTIVE.has(previous) && TERMINAL.has(job.status)) notifyTerminal(job);
+      known.set(job.job_id, job.status);
+    });
+    jobs = next;
+    render();
+  } catch (_) { /* 后台刷新失败静默，下轮恢复 */ }
+  schedule();
+}
+
+function schedule() {
+  if (timer) clearTimeout(timer);
+  const delay = jobs.some((job) => ACTIVE.has(job.status)) ? POLL_ACTIVE_MS : POLL_IDLE_MS;
+  timer = setTimeout(refreshSoftwareJobs, document.hidden ? Math.max(delay, 30000) : delay);
+}
+
+document.addEventListener("visibilitychange", () => {
+  if (!document.hidden && state.token) refreshSoftwareJobs(); else schedule();
+});
+
+function render() {
+  const center = $("software-job-center");
+  const panel = $("software-job-panel");
+  const active = jobs.filter((job) => ACTIVE.has(job.status));
+  const recent = jobs.slice(0, 10);
+  center.classList.toggle("show", active.length > 0 || recent.length > 0);
+  center.classList.toggle("expanded", expanded);
+  $("software-job-count").textContent = active.length ? `${active.length} 运行中` : `${jobs.length} 条`;
+  panel.hidden = !expanded;
+  if (!expanded) return;
+  panel.innerHTML = recent.length
+    ? recent.map(jobCard).join("")
+    : '<div class="sj-empty">暂无专业软件任务</div>';
+  panel.querySelectorAll("[data-job-action]").forEach((button) => {
+    button.onclick = (event) => handleAction(event, button);
+  });
+}
+
+function jobCard(job) {
+  const summary = job.request_summary || {};
+  const input = job.input || {};
+  const active = ACTIVE.has(job.status);
+  const progress = Math.max(0, Math.min(100, Number(job.progress || 0)));
+  const detail = stageLabel[job.stage] || statusLabel[job.status] || "正在处理";
+  const error = job.error && (job.error.detail || job.error.code);
+  return `<article class="sj-card ${escapeHtml(job.status)}" data-job-id="${escapeHtml(job.job_id)}">
+    <div class="sj-title"><strong>${escapeHtml(summary.display_name || job.capability)}</strong>
+      <span>${escapeHtml(statusLabel[job.status] || job.status)}</span></div>
+    <div class="sj-sub">${escapeHtml(detail)}${error ? ` · ${escapeHtml(error)}` : ""}</div>
+    ${active ? `<div class="sj-progress"><i style="width:${progress}%"></i></div>` : ""}
+    <div class="sj-meta">${escapeHtml(job.task_name || "未命名对话")}${input.filename ? ` · ${escapeHtml(input.filename)}` : ""}</div>
+    <div class="sj-actions">
+      <button class="small" data-job-action="open" data-task-id="${escapeHtml(job.task_id)}">打开对话</button>
+      ${job.status === "succeeded" ? `<button class="small primary" data-job-action="analyze" data-task-id="${escapeHtml(job.task_id)}">分析结果</button>` : ""}
+      ${active && job.status !== "cancelling" ? '<button class="small danger" data-job-action="cancel">停止</button>' : ""}
+    </div>
+  </article>`;
+}
+
+async function handleAction(event, button) {
+  event.stopPropagation();
+  const card = button.closest("[data-job-id]");
+  const jobId = card.dataset.jobId;
+  const action = button.dataset.jobAction;
+  if (action === "cancel") {
+    if (!await dialogConfirm({
+      title: "停止专业软件任务",
+      message: "确定停止这个任务？已经产生但尚未发布的中间输出可能不会保留。",
+      okText: "停止", danger: true,
+    })) return;
+    button.disabled = true;
+    try { await api("POST", `/v1/software-jobs/${jobId}/cancel`); }
+    catch (error) { message(error.message || "停止失败", "error"); }
+    refreshSoftwareJobs();
+    return;
+  }
+  const taskId = button.dataset.taskId;
+  if (taskId) await selectTask(taskId);
+  expanded = false;
+  render();
+  if (action === "analyze") {
+    setTimeout(() => {
+      const input = $("chat-input");
+      if (!input) return;
+      input.value = `请分析专业软件任务 ${jobId} 的结果，结合输出图表和输入数据总结主要结论。`;
+      input.focus();
+      input.dispatchEvent(new Event("input", { bubbles: true }));
+    }, 400);
+  }
+}
+
+function notifyTerminal(job) {
+  const ok = job.status === "succeeded";
+  const label = ok ? "已完成" : (job.status === "cancelled" ? "已取消" : "失败");
+  const summary = job.request_summary || {};
+  message(`${summary.display_name || "专业软件任务"}${label}`, ok ? "success" : "error", 6000);
+  expanded = true;
+}

windows-node/README.md

@@ -2,7 +2,15 @@
 
 内网 MVP 的 Windows 执行节点，目标运行环境为 Windows 11 Enterprise + .NET 10 SDK 10.0.303。仓库根目录 `global.json` 固定 SDK patch；客户端只使用 .NET Windows Desktop Framework，不依赖第三方 NuGet 包。
 
-当前实现托盘状态角标、小型配置窗口、注册、DPAPI/ACL 配置保存、WebSocket `hello`/心跳和退避重连。尚未实现 `compute_jobs`、Origin Worker、任务目录和产物上传。
+当前实现托盘状态角标、小型配置窗口、注册、DPAPI/ACL 配置保存、WebSocket `hello`/心跳和退避重连，并只读探测 Origin/OriginPro 安装版本、COM 自动化组件与桌面会话状态。Node 可以接收受控的 `origin.plot@v1` offer，在本机任务目录原子保存请求后回报 accept/reject；随后以 Node 身份流式下载任务绑定的 CSV/XLSX/JSON，校验大小与 SHA-256 后原子保存。固定 Origin Worker 独立于单次 WebSocket 执行，断线不终止已启动绘图。成功产物按 manifest 逐项流式上传并由云端复核大小与 SHA-256，全部完成后原子发布到任务工作目录的 `origin/<job_id>/`；中断后按本地 `upload-complete.json` 幂等续传。
+
+固定 Origin Worker 已支持 `line`、`scatter` 和 `line_scatter`，生成 OPJU、PNG、SVG、PDF、plot spec、provenance 与原子 `terminal.json`。运行时独立于 zcbot 服务端 Python，管理员执行：
+
+```powershell
+.\install-origin-runtime.ps1 -BootstrapPython D:\programs\Python312\python.exe
+```
+
+默认安装到 `%ProgramData%\Zcbot\WindowsNode\runtimes\origin\python.exe`；也可由管理员用绝对路径环境变量 `ZCBOT_ORIGIN_PYTHON` 指向固定解释器。任务请求不能指定解释器、脚本或路径。
 
 注册和运行必须使用同一专用 Windows 账号。MVP 通过该账号的登录后计划任务自动启动，不以 Windows Service 在 Session 0 运行。
 

windows-node/Zcbot.WindowsNode/ConfigurationForm.cs

@@ -2,12 +2,13 @@ namespace Zcbot.WindowsNode;
 
 internal sealed class ConfigurationForm : Form
 {
-    private const int ContentWidth = 800;
+    private const int ContentWidth = 760;
 
     private readonly TextBox server = CreateTextBox("http://127.0.0.1:8765");
     private readonly TextBox nodeName = CreateTextBox(Environment.MachineName.ToLowerInvariant());
     private readonly TextBox enrollmentCode = CreateTextBox(usePassword: true);
     private readonly Button register = CreateButton("注册并连接", 128, primary: true);
+    private readonly Button reconnect = CreateButton("立即重连", 112, primary: true);
     private readonly Button resetIdentity = CreateButton("清除本机身份并重新注册", 220);
     private readonly CheckBox startAtLogin = new()
     {
@@ -25,18 +26,18 @@ internal sealed class ConfigurationForm : Form
     private readonly Label identity = CreateBodyLabel();
     private readonly Label capabilitySummary = CreateBodyLabel();
     private readonly TableLayoutPanel registrationCard;
-    private readonly TableLayoutPanel registeredActionsCard;
     private bool changingStartup;
 
     internal event Func<EnrollOptions, Task>? RegisterRequested;
+    internal event Action? ReconnectRequested;
     internal event Action? ResetIdentityRequested;
 
     internal ConfigurationForm()
     {
         Text = "zcbot Windows Node";
         AutoScaleMode = AutoScaleMode.Dpi;
-        ClientSize = new Size(880, 720);
-        MinimumSize = new Size(850, 680);
+        ClientSize = new Size(840, 680);
+        MinimumSize = new Size(820, 640);
         StartPosition = FormStartPosition.CenterScreen;
         Font = new Font("Microsoft YaHei UI", 9);
         FormBorderStyle = FormBorderStyle.Sizable;
@@ -62,7 +63,7 @@ internal sealed class ConfigurationForm : Form
             AutoSizeMode = AutoSizeMode.GrowAndShrink,
             Dock = DockStyle.Top,
             ColumnCount = 1,
-            RowCount = 6,
+            RowCount = 5,
             BackColor = BackColor,
         };
         page.ColumnStyles.Add(new ColumnStyle(SizeType.Percent, 100));
@@ -85,33 +86,19 @@ internal sealed class ConfigurationForm : Form
         heading.Controls.Add(CreateHint("连接本机科研软件与 zcbot 的受控执行节点"));
         page.Controls.Add(heading);
 
-        var overview = new TableLayoutPanel
-        {
-            AutoSize = true,
-            Dock = DockStyle.Top,
-            ColumnCount = 2,
-            Margin = new Padding(0, 0, 0, 14),
-        };
-        overview.ColumnStyles.Add(new ColumnStyle(SizeType.Percent, 50));
-        overview.ColumnStyles.Add(new ColumnStyle(SizeType.Percent, 50));
-
         var statusCard = CreateCard();
-        statusCard.Margin = new Padding(0, 0, 7, 0);
-        statusCard.Controls.Add(CreateSectionTitle("节点状态"));
+        statusCard.Controls.Add(CreateSectionTitle("节点"));
         statusCard.Controls.Add(state);
         statusCard.Controls.Add(detail);
         statusCard.Controls.Add(identity);
-        overview.Controls.Add(statusCard, 0, 0);
-
-        var capabilityCard = CreateCard();
-        capabilityCard.Margin = new Padding(7, 0, 0, 0);
-        capabilityCard.Controls.Add(CreateSectionTitle("节点能力"));
-        capabilityCard.Controls.Add(CreateCapabilityBadge("Origin 绘图", "origin.plot@v1"));
-        capabilityCard.Controls.Add(capabilitySummary);
-        capabilityCard.Controls.Add(CreateHint(
-            "当前 MVP 内置声明该协议，不需要手工配置；它不代表已完成 Origin 安装检测。"));
-        overview.Controls.Add(capabilityCard, 1, 0);
-        page.Controls.Add(overview);
+        statusCard.Controls.Add(CreateDivider());
+        statusCard.Controls.Add(CreateCapabilityRow("Origin 绘图", "origin.plot@v1"));
+        statusCard.Controls.Add(capabilitySummary);
+        var resetActions = CreateActions();
+        resetActions.Controls.Add(reconnect);
+        resetActions.Controls.Add(resetIdentity);
+        statusCard.Controls.Add(resetActions);
+        page.Controls.Add(statusCard);
 
         registrationCard = CreateCard();
         registrationCard.Controls.Add(CreateSectionTitle("首次注册"));
@@ -125,15 +112,6 @@ internal sealed class ConfigurationForm : Form
         registrationCard.Controls.Add(registerActions);
         page.Controls.Add(registrationCard);
 
-        registeredActionsCard = CreateCard();
-        registeredActionsCard.Controls.Add(CreateSectionTitle("节点身份"));
-        registeredActionsCard.Controls.Add(CreateHint(
-            "服务地址、节点名称和能力在首次注册时确定。需要修改时，请先在管理后台删除旧节点，再清除本机身份并重新注册。"));
-        var identityActions = CreateActions();
-        identityActions.Controls.Add(resetIdentity);
-        registeredActionsCard.Controls.Add(identityActions);
-        page.Controls.Add(registeredActionsCard);
-
         var runtimeCard = CreateCard();
         runtimeCard.Controls.Add(CreateSectionTitle("运行设置"));
         runtimeCard.Controls.Add(startAtLogin);
@@ -147,6 +125,7 @@ internal sealed class ConfigurationForm : Form
         page.Controls.Add(securityNote);
 
         register.Click += async (_, _) => await RegisterAsync();
+        reconnect.Click += (_, _) => ReconnectRequested?.Invoke();
         resetIdentity.Click += (_, _) => ResetIdentity();
         startAtLogin.CheckedChanged += (_, _) => ToggleStartup();
         FormClosing += (_, eventArgs) =>
@@ -185,11 +164,12 @@ internal sealed class ConfigurationForm : Form
             : $"节点：{config.NodeName}\nNode ID：{config.NodeId}\n服务：{config.ServerUrl}";
         capabilitySummary.Text = config is null
             ? "注册后启用"
-            : $"已声明 {config.Capabilities.Count} 项能力";
+            : FormatOriginStatus(OriginRuntimeProbe.Detect());
 
         var registered = config is not null;
         registrationCard.Visible = !registered;
-        registeredActionsCard.Visible = registered;
+        reconnect.Visible = registered;
+        resetIdentity.Visible = registered;
         register.Enabled = !registered && status.State != NodeState.Connecting;
         if (registered)
         {
@@ -199,6 +179,15 @@ internal sealed class ConfigurationForm : Form
         }
     }
 
+    private static string FormatOriginStatus(OriginRuntimeStatus origin)
+    {
+        var version = string.IsNullOrWhiteSpace(origin.SoftwareVersion)
+            ? "版本未知"
+            : $"版本 {origin.SoftwareVersion}";
+        var state = origin.Health == "ready" ? "可用" : "不可用";
+        return $"{state} · {version}\n{origin.Detail}";
+    }
+
     private void ResetIdentity()
     {
         var answer = MessageBox.Show(
@@ -287,7 +276,7 @@ internal sealed class ConfigurationForm : Form
         AutoSize = true,
         Dock = DockStyle.Fill,
         ForeColor = Color.FromArgb(71, 85, 105),
-        MaximumSize = new Size(340, 0),
+        MaximumSize = new Size(ContentWidth - 40, 0),
         Margin = new Padding(0, 2, 0, 5),
     };
 
@@ -300,25 +289,33 @@ internal sealed class ConfigurationForm : Form
         Margin = new Padding(0, 2, 0, 4),
     };
 
-    private static Panel CreateCapabilityBadge(string title, string protocol)
+    private static Panel CreateDivider() => new()
     {
-        var badge = new Panel
+        Height = 1,
+        Dock = DockStyle.Top,
+        BackColor = Color.FromArgb(226, 232, 240),
+        Margin = new Padding(0, 10, 0, 12),
+    };
+
+    private static Panel CreateCapabilityRow(string title, string protocol)
+    {
+        var row = new Panel
         {
             AutoSize = false,
-            Height = 54,
+            Height = 36,
             Dock = DockStyle.Top,
             BackColor = Color.FromArgb(239, 246, 255),
-            Margin = new Padding(0, 2, 0, 8),
-            Padding = new Padding(12, 7, 12, 7),
+            Margin = new Padding(0, 0, 0, 5),
+            Padding = new Padding(11, 7, 11, 7),
         };
-        badge.Controls.Add(new Label
+        row.Controls.Add(new Label
         {
-            Text = $"{title}\n协议：{protocol}",
+            Text = $"{title}　·　{protocol}",
             AutoSize = true,
             Font = new Font("Microsoft YaHei UI", 9, FontStyle.Bold),
             ForeColor = Color.FromArgb(29, 78, 216),
         });
-        return badge;
+        return row;
     }
 
     private static TextBox CreateTextBox(string text = "", bool usePassword = false) => new()

windows-node/Zcbot.WindowsNode/EnrollOptions.cs

@@ -52,7 +52,7 @@ internal static class NodeUri
 
     internal static Uri WebSocketEndpoint(Uri serverUrl)
     {
-        var builder = new UriBuilder(new Uri(serverUrl, "v1/compute/nodes/connect"))
+        var builder = new UriBuilder(new Uri(serverUrl, "v1/software-nodes/connect"))
         {
             Scheme = serverUrl.Scheme == Uri.UriSchemeHttps ? "wss" : "ws"
         };

windows-node/Zcbot.WindowsNode/EnrollmentClient.cs

@@ -28,7 +28,7 @@ internal static class EnrollmentClient
 
         using var client = new HttpClient { BaseAddress = options.ServerUrl, Timeout = TimeSpan.FromSeconds(30) };
         using var response = await client.PostAsJsonAsync(
-            "v1/compute/nodes/enroll", request, cancellationToken);
+            "v1/software-nodes/enroll", request, cancellationToken);
         if (!response.IsSuccessStatusCode)
         {
             var detail = await response.Content.ReadAsStringAsync(cancellationToken);

windows-node/Zcbot.WindowsNode/JobInboxStore.cs

@@ -0,0 +1,346 @@
+using System.Text;
+using System.Text.Json;
+
+namespace Zcbot.WindowsNode;
+
+internal sealed class JobInboxStore(string jobsDirectory)
+{
+    private static readonly JsonSerializerOptions JsonOptions = new() { WriteIndented = true };
+    private static readonly HashSet<string> PlotTypes =
+    ["line", "scatter", "line_scatter"];
+    private static readonly HashSet<string> OutputFormats = ["opju", "png", "svg", "pdf"];
+
+    internal bool HasPendingJobs => Directory.Exists(jobsDirectory)
+        && ReadRecoverableJobs().Any(item =>
+            item.Terminal is null
+            || item.Terminal.Value.GetProperty("status").GetString() == "succeeded"
+                && !item.UploadComplete);
+
+    internal IReadOnlyList<RecoverableJob> ReadRecoverableJobs()
+    {
+        if (!Directory.Exists(jobsDirectory))
+        {
+            return [];
+        }
+        var jobs = new List<RecoverableJob>();
+        foreach (var requestPath in Directory.EnumerateFiles(
+            jobsDirectory, "request.json", SearchOption.AllDirectories))
+        {
+            try
+            {
+                using var request = JsonDocument.Parse(File.ReadAllBytes(requestPath));
+                var root = request.RootElement;
+                if (!TryReadGuid(root, "job_id", out var jobId)
+                    || !TryReadGuid(root, "lease_id", out var leaseId)
+                    || !root.TryGetProperty("request_digest", out var digestValue)
+                    || digestValue.GetString() is not { Length: 64 } requestDigest)
+                {
+                    continue;
+                }
+                var jobDirectory = Directory.GetParent(Directory.GetParent(requestPath)!.FullName)!.FullName;
+                jobs.Add(new RecoverableJob(
+                    jobId,
+                    leaseId,
+                    requestDigest,
+                    root.TryGetProperty("input_transfer", out var transfer)
+                        ? transfer.Clone() : null,
+                    ReadTerminal(Path.Combine(jobDirectory, "terminal.json")),
+                    File.Exists(Path.Combine(jobDirectory, "upload-complete.json"))));
+            }
+            catch (Exception exception) when (
+                exception is JsonException or IOException or UnauthorizedAccessException)
+            {
+            }
+        }
+        return jobs;
+    }
+
+    internal JobOfferResult Accept(JsonElement payload)
+    {
+        if (!TryReadGuid(payload, "job_id", out var jobId)
+            || !TryReadGuid(payload, "lease_id", out var leaseId)
+            || !payload.TryGetProperty("request_digest", out var digestValue)
+            || digestValue.ValueKind != JsonValueKind.String
+            || digestValue.GetString() is not { Length: 64 } requestDigest
+            || !payload.TryGetProperty("capability", out var capabilityValue)
+            || capabilityValue.GetString() != "origin.plot@v1"
+            || !payload.TryGetProperty("request", out var request)
+            || request.ValueKind != JsonValueKind.Object
+            || !payload.TryGetProperty("input_transfer", out var inputTransfer)
+            || !IsValidInputTransfer(inputTransfer))
+        {
+            return JobOfferResult.Reject("invalid_offer");
+        }
+        if (!IsValidRequest(request))
+        {
+            return JobOfferResult.Reject("unsupported_request");
+        }
+
+        var directory = Path.Combine(jobsDirectory, jobId.ToString("D"));
+        var requestDirectory = Path.Combine(directory, "request");
+        var requestPath = Path.Combine(requestDirectory, "request.json");
+        Directory.CreateDirectory(requestDirectory);
+        if (File.Exists(requestPath))
+        {
+            try
+            {
+                using var existing = JsonDocument.Parse(File.ReadAllBytes(requestPath));
+                var root = existing.RootElement;
+                var sameDigest = root.TryGetProperty("request_digest", out var existingDigest)
+                    && existingDigest.GetString() == requestDigest;
+                if (!sameDigest)
+                {
+                    return JobOfferResult.Reject("job_digest_conflict");
+                }
+                if (!TryReadGuid(root, "lease_id", out var existingLease)
+                    || existingLease != leaseId)
+                {
+                    var updated = JsonSerializer.SerializeToUtf8Bytes(new
+                    {
+                        job_id = jobId,
+                        lease_id = leaseId,
+                        request_digest = requestDigest,
+                        capability = "origin.plot@v1",
+                        accepted_at = DateTimeOffset.UtcNow,
+                        request,
+                        input_transfer = inputTransfer,
+                    }, JsonOptions);
+                    AtomicWrite(requestPath, updated, overwrite: true);
+                }
+                return JobOfferResult.Accept(jobId, leaseId, requestDigest);
+            }
+            catch (JsonException)
+            {
+                return JobOfferResult.Reject("local_job_record_invalid");
+            }
+        }
+
+        var record = JsonSerializer.SerializeToUtf8Bytes(new
+        {
+            job_id = jobId,
+            lease_id = leaseId,
+            request_digest = requestDigest,
+            capability = "origin.plot@v1",
+            accepted_at = DateTimeOffset.UtcNow,
+            request,
+            input_transfer = inputTransfer,
+        }, JsonOptions);
+        try
+        {
+            AtomicWrite(requestPath, record, overwrite: false);
+            return JobOfferResult.Accept(jobId, leaseId, requestDigest);
+        }
+        catch (IOException)
+        {
+            return JobOfferResult.Reject("local_job_persist_failed");
+        }
+    }
+
+    private static JsonElement? ReadTerminal(string path)
+    {
+        if (!File.Exists(path))
+        {
+            return null;
+        }
+        using var document = JsonDocument.Parse(File.ReadAllBytes(path));
+        return document.RootElement.Clone();
+    }
+
+    private static void AtomicWrite(string path, byte[] content, bool overwrite)
+    {
+        Directory.CreateDirectory(Path.GetDirectoryName(path)!);
+        var temporaryPath = path + ".tmp-" + Guid.NewGuid().ToString("N");
+        try
+        {
+            using (var stream = new FileStream(
+                temporaryPath, FileMode.CreateNew, FileAccess.Write, FileShare.None,
+                bufferSize: 4096, FileOptions.WriteThrough))
+            {
+                stream.Write(content);
+                stream.Flush(flushToDisk: true);
+            }
+            File.Move(temporaryPath, path, overwrite);
+        }
+        finally
+        {
+            if (File.Exists(temporaryPath)) File.Delete(temporaryPath);
+        }
+    }
+
+    private static bool IsValidRequest(JsonElement request) =>
+        HasOnlyProperties(request, "schema_version", "input", "plot", "output")
+        && request.TryGetProperty("schema_version", out var schemaVersion)
+        && schemaVersion.TryGetInt32(out var version)
+        && version == 1
+        && request.TryGetProperty("input", out var input)
+        && input.ValueKind == JsonValueKind.Object
+        && HasOnlyProperties(input, "input_id", "sheet")
+        && input.TryGetProperty("input_id", out var inputId)
+        && inputId.ValueKind == JsonValueKind.String
+        && !string.IsNullOrWhiteSpace(inputId.GetString())
+        && request.TryGetProperty("plot", out var plot)
+        && plot.ValueKind == JsonValueKind.Object
+        && HasOnlyProperties(
+            plot, "type", "x", "y", "template", "title", "x_axis", "y_axis", "legend", "error_bars")
+        && plot.TryGetProperty("type", out var plotType)
+        && plotType.ValueKind == JsonValueKind.String
+        && PlotTypes.Contains(plotType.GetString() ?? "")
+        && (!plot.TryGetProperty("title", out var title)
+            || title.ValueKind == JsonValueKind.String && title.GetString()!.Length <= 500)
+        && plot.TryGetProperty("x", out var x)
+        && IsColumnName(x)
+        && plot.TryGetProperty("y", out var y)
+        && IsValidYColumns(y)
+        && (!plot.TryGetProperty("template", out var template)
+            || template.GetString() == "publication_double_column")
+        && IsValidAxis(plot, "x_axis")
+        && IsValidAxis(plot, "y_axis")
+        && IsValidLegend(plot)
+        && !plot.TryGetProperty("error_bars", out _)
+        && request.TryGetProperty("output", out var output)
+        && output.ValueKind == JsonValueKind.Object
+        && HasOnlyProperties(output, "formats", "dpi", "capture_screenshots", "record_video")
+        && output.TryGetProperty("formats", out var formats)
+        && formats.ValueKind == JsonValueKind.Array
+        && formats.GetArrayLength() > 0
+        && IsValidFormats(formats)
+        && (!output.TryGetProperty("dpi", out var dpi)
+            || dpi.TryGetInt32(out var dpiValue) && dpiValue is >= 72 and <= 1200)
+        && IsOptionalBoolean(output, "capture_screenshots")
+        && IsOptionalBoolean(output, "record_video")
+        && (!output.TryGetProperty("record_video", out var recordVideo)
+            || recordVideo.ValueKind == JsonValueKind.False);
+
+    private static bool IsColumnName(JsonElement value) =>
+        value.ValueKind == JsonValueKind.String
+        && value.GetString() is { Length: >= 1 and <= 128 };
+
+    private static bool IsValidFormats(JsonElement formats)
+    {
+        var values = formats.EnumerateArray().ToArray();
+        return values.All(item =>
+                item.ValueKind == JsonValueKind.String
+                && OutputFormats.Contains(item.GetString() ?? ""))
+            && values.Select(item => item.GetString()).Distinct(StringComparer.Ordinal).Count()
+                == values.Length;
+    }
+
+    private static bool IsValidYColumns(JsonElement value)
+    {
+        if (IsColumnName(value)) return true;
+        if (value.ValueKind != JsonValueKind.Array
+            || value.GetArrayLength() is < 1 or > 16)
+        {
+            return false;
+        }
+        var names = value.EnumerateArray().Select(item => item.GetString()).ToArray();
+        return value.EnumerateArray().All(IsColumnName)
+            && names.Distinct(StringComparer.Ordinal).Count() == names.Length;
+    }
+
+    private static bool IsValidAxis(JsonElement plot, string name)
+    {
+        if (!plot.TryGetProperty(name, out var axis)) return true;
+        return axis.ValueKind == JsonValueKind.Object
+            && HasOnlyProperties(axis, "title", "unit", "scale")
+            && (!axis.TryGetProperty("title", out var title) || title.ValueKind == JsonValueKind.String)
+            && (!axis.TryGetProperty("unit", out var unit) || unit.ValueKind == JsonValueKind.String)
+            && (!axis.TryGetProperty("scale", out var scale) || scale.GetString() == "linear");
+    }
+
+    private static bool IsValidLegend(JsonElement plot)
+    {
+        if (!plot.TryGetProperty("legend", out var legend)) return true;
+        return legend.ValueKind == JsonValueKind.Object
+            && HasOnlyProperties(legend, "enabled", "position")
+            && (!legend.TryGetProperty("enabled", out var enabled)
+                || enabled.ValueKind == JsonValueKind.True)
+            && (!legend.TryGetProperty("position", out var position)
+                || position.GetString() == "top_right");
+    }
+
+    private static bool IsOptionalBoolean(JsonElement value, string name) =>
+        !value.TryGetProperty(name, out var property)
+        || property.ValueKind is JsonValueKind.True or JsonValueKind.False;
+
+    private static bool IsValidInputTransfer(JsonElement transfer) =>
+        transfer.ValueKind == JsonValueKind.Object
+        && HasOnlyProperties(transfer, "artifact_id", "filename", "size_bytes", "sha256", "download_path")
+        && transfer.TryGetProperty("artifact_id", out var artifactId)
+        && Guid.TryParse(artifactId.GetString(), out _)
+        && transfer.TryGetProperty("filename", out var filename)
+        && filename.ValueKind == JsonValueKind.String
+        && Path.GetFileName(filename.GetString()) == filename.GetString()
+        && transfer.TryGetProperty("size_bytes", out var size)
+        && size.TryGetInt64(out var sizeBytes)
+        && sizeBytes is >= 0 and <= 104_857_600
+        && transfer.TryGetProperty("sha256", out var sha)
+        && sha.GetString() is { Length: 64 }
+        && transfer.TryGetProperty("download_path", out var downloadPath)
+        && downloadPath.GetString()?.StartsWith("/v1/software-jobs/", StringComparison.Ordinal) == true
+        && downloadPath.GetString()?.EndsWith("/input", StringComparison.Ordinal) == true;
+
+    internal string InputPath(RecoverableJob job)
+    {
+        if (job.InputTransfer is not JsonElement transfer
+            || !IsValidInputTransfer(transfer))
+        {
+            throw new InvalidDataException("Stored input transfer is invalid.");
+        }
+        var filename = transfer.GetProperty("filename").GetString()!;
+        return Path.Combine(jobsDirectory, job.JobId.ToString("D"), "input", filename);
+    }
+
+    internal void WriteTerminal(
+        RecoverableJob job, string status, string code, string detail)
+    {
+        var path = Path.Combine(jobsDirectory, job.JobId.ToString("D"), "terminal.json");
+        if (File.Exists(path))
+        {
+            return;
+        }
+        var content = JsonSerializer.SerializeToUtf8Bytes(new
+        {
+            job_id = job.JobId,
+            lease_id = job.LeaseId,
+            request_digest = job.RequestDigest,
+            status,
+            error = new { code, detail },
+            artifact_manifest = Array.Empty<object>(),
+            terminal_at = DateTimeOffset.UtcNow,
+        }, JsonOptions);
+        AtomicWrite(path, content, overwrite: false);
+    }
+
+    private static bool HasOnlyProperties(JsonElement value, params string[] allowed)
+    {
+        var names = new HashSet<string>(allowed, StringComparer.Ordinal);
+        return value.EnumerateObject().All(item => names.Contains(item.Name));
+    }
+
+    private static bool TryReadGuid(JsonElement payload, string name, out Guid value)
+    {
+        value = Guid.Empty;
+        return payload.TryGetProperty(name, out var property)
+            && property.ValueKind == JsonValueKind.String
+            && Guid.TryParse(property.GetString(), out value);
+    }
+}
+
+internal sealed record RecoverableJob(
+    Guid JobId,
+    Guid LeaseId,
+    string RequestDigest,
+    JsonElement? InputTransfer,
+    JsonElement? Terminal,
+    bool UploadComplete);
+
+internal sealed record JobOfferResult(
+    bool Accepted, Guid JobId, Guid LeaseId, string RequestDigest, string Reason)
+{
+    internal static JobOfferResult Accept(Guid jobId, Guid leaseId, string requestDigest) =>
+        new(true, jobId, leaseId, requestDigest, "");
+
+    internal static JobOfferResult Reject(string reason) =>
+        new(false, Guid.Empty, Guid.Empty, "", reason);
+}

windows-node/Zcbot.WindowsNode/JobInputDownloader.cs

@@ -0,0 +1,99 @@
+using System.Net.Http.Headers;
+using System.Security.Cryptography;
+using System.Text.Json;
+
+namespace Zcbot.WindowsNode;
+
+internal sealed class JobInputDownloader(NodeConfig config, JobInboxStore inbox)
+{
+    internal async Task DownloadAsync(RecoverableJob job, CancellationToken cancellationToken)
+    {
+        if (job.InputTransfer is not JsonElement transfer)
+        {
+            throw new InvalidDataException("Job input transfer is missing.");
+        }
+        var downloadPath = transfer.GetProperty("download_path").GetString()!;
+        if (!downloadPath.StartsWith("/v1/software-jobs/", StringComparison.Ordinal)
+            || !downloadPath.EndsWith("/input", StringComparison.Ordinal)
+            || !Uri.TryCreate(downloadPath, UriKind.Relative, out var relativeUri))
+        {
+            throw new InvalidDataException("Job input download path is invalid.");
+        }
+        var expectedSize = transfer.GetProperty("size_bytes").GetInt64();
+        var expectedSha256 = transfer.GetProperty("sha256").GetString()!;
+        var destination = inbox.InputPath(job);
+        if (File.Exists(destination))
+        {
+            await VerifyExistingAsync(destination, expectedSize, expectedSha256, cancellationToken);
+            return;
+        }
+
+        Directory.CreateDirectory(Path.GetDirectoryName(destination)!);
+        var temporaryPath = destination + ".tmp-" + Guid.NewGuid().ToString("N");
+        using var client = new HttpClient { BaseAddress = config.ServerUrl };
+        client.DefaultRequestHeaders.Authorization =
+            new AuthenticationHeaderValue("Bearer", config.NodeToken);
+        client.DefaultRequestHeaders.Add("X-Node-Id", config.NodeId.ToString());
+        try
+        {
+            using var response = await client.GetAsync(
+                relativeUri, HttpCompletionOption.ResponseHeadersRead, cancellationToken);
+            response.EnsureSuccessStatusCode();
+            if (response.Content.Headers.ContentLength is long contentLength
+                && contentLength != expectedSize)
+            {
+                throw new InvalidDataException("Job input size header does not match the manifest.");
+            }
+            await using var source = await response.Content.ReadAsStreamAsync(cancellationToken);
+            await using var target = new FileStream(
+                temporaryPath, FileMode.CreateNew, FileAccess.Write, FileShare.None,
+                bufferSize: 64 * 1024, FileOptions.Asynchronous | FileOptions.WriteThrough);
+            using var hash = IncrementalHash.CreateHash(HashAlgorithmName.SHA256);
+            var buffer = new byte[64 * 1024];
+            long total = 0;
+            while (true)
+            {
+                var count = await source.ReadAsync(buffer, cancellationToken);
+                if (count == 0) break;
+                total += count;
+                if (total > expectedSize)
+                {
+                    throw new InvalidDataException("Job input exceeded its declared size.");
+                }
+                hash.AppendData(buffer, 0, count);
+                await target.WriteAsync(buffer.AsMemory(0, count), cancellationToken);
+            }
+            await target.FlushAsync(cancellationToken);
+            target.Flush(flushToDisk: true);
+            var actualSha256 = Convert.ToHexString(hash.GetHashAndReset()).ToLowerInvariant();
+            if (total != expectedSize || actualSha256 != expectedSha256)
+            {
+                throw new InvalidDataException("Job input digest does not match the manifest.");
+            }
+            target.Close();
+            File.Move(temporaryPath, destination, overwrite: false);
+        }
+        finally
+        {
+            if (File.Exists(temporaryPath)) File.Delete(temporaryPath);
+        }
+    }
+
+    private static async Task VerifyExistingAsync(
+        string path, long expectedSize, string expectedSha256, CancellationToken cancellationToken)
+    {
+        var info = new FileInfo(path);
+        if (info.Length != expectedSize)
+        {
+            throw new InvalidDataException("Existing job input size does not match the manifest.");
+        }
+        await using var stream = new FileStream(
+            path, FileMode.Open, FileAccess.Read, FileShare.Read, 64 * 1024, FileOptions.Asynchronous);
+        var digest = Convert.ToHexString(
+            await SHA256.HashDataAsync(stream, cancellationToken)).ToLowerInvariant();
+        if (digest != expectedSha256)
+        {
+            throw new InvalidDataException("Existing job input digest does not match the manifest.");
+        }
+    }
+}

windows-node/Zcbot.WindowsNode/JobOutputUploader.cs

@@ -0,0 +1,100 @@
+using System.Net.Http.Headers;
+using System.Security.Cryptography;
+using System.Text;
+using System.Text.Json;
+
+namespace Zcbot.WindowsNode;
+
+internal sealed class JobOutputUploader(NodeConfig config)
+{
+    private static readonly JsonSerializerOptions JsonOptions = new() { WriteIndented = true };
+
+    internal async Task UploadAsync(RecoverableJob job)
+    {
+        var jobDirectory = Path.Combine(
+            NodePaths.ForCurrentMachine().JobsDirectory, job.JobId.ToString("D"));
+        var completionPath = Path.Combine(jobDirectory, "upload-complete.json");
+        if (File.Exists(completionPath)) return;
+        var terminalPath = Path.Combine(jobDirectory, "terminal.json");
+        using var terminal = JsonDocument.Parse(await File.ReadAllBytesAsync(terminalPath));
+        if (terminal.RootElement.GetProperty("status").GetString() != "succeeded") return;
+        var manifest = terminal.RootElement.GetProperty("artifact_manifest").Clone();
+
+        using var client = new HttpClient { BaseAddress = config.ServerUrl };
+        client.DefaultRequestHeaders.Authorization =
+            new AuthenticationHeaderValue("Bearer", config.NodeToken);
+        client.DefaultRequestHeaders.Add("X-Node-Id", config.NodeId.ToString());
+        client.DefaultRequestHeaders.Add("X-Lease-Id", job.LeaseId.ToString());
+        client.DefaultRequestHeaders.Add("X-Request-Digest", job.RequestDigest);
+        foreach (var artifact in manifest.EnumerateArray())
+        {
+            var localId = artifact.GetProperty("artifact_id").GetString()!;
+            var filename = artifact.GetProperty("filename").GetString()!;
+            var expectedSize = artifact.GetProperty("size_bytes").GetInt64();
+            var expectedDigest = artifact.GetProperty("sha256").GetString()!;
+            var path = Path.Combine(jobDirectory, "output", filename);
+            var info = new FileInfo(path);
+            if (!info.Exists || info.Length != expectedSize)
+            {
+                throw new InvalidDataException($"Output artifact is missing or changed: {localId}.");
+            }
+            await using (var verify = new FileStream(
+                path, FileMode.Open, FileAccess.Read, FileShare.Read, 64 * 1024,
+                FileOptions.Asynchronous | FileOptions.SequentialScan))
+            {
+                var digest = Convert.ToHexString(
+                    await SHA256.HashDataAsync(verify)).ToLowerInvariant();
+                if (digest != expectedDigest)
+                {
+                    throw new InvalidDataException($"Output artifact digest changed: {localId}.");
+                }
+            }
+            await using var stream = new FileStream(
+                path, FileMode.Open, FileAccess.Read, FileShare.Read, 64 * 1024,
+                FileOptions.Asynchronous | FileOptions.SequentialScan);
+            using var content = new StreamContent(stream);
+            content.Headers.ContentType = new MediaTypeHeaderValue("application/octet-stream");
+            content.Headers.ContentLength = expectedSize;
+            content.Headers.Add("X-Content-SHA256", expectedDigest);
+            content.Headers.Add("X-Content-Length", expectedSize.ToString());
+            using var response = await client.PutAsync(
+                $"/v1/software-jobs/{job.JobId:D}/outputs/{Uri.EscapeDataString(localId)}",
+                content);
+            response.EnsureSuccessStatusCode();
+        }
+
+        using var completeContent = new StringContent(
+            JsonSerializer.Serialize(new { artifact_manifest = manifest }),
+            Encoding.UTF8,
+            "application/json");
+        using var completeResponse = await client.PostAsync(
+            $"/v1/software-jobs/{job.JobId:D}/outputs/complete", completeContent);
+        completeResponse.EnsureSuccessStatusCode();
+        var responseBody = await completeResponse.Content.ReadAsByteArrayAsync();
+        AtomicWrite(completionPath, responseBody);
+    }
+
+    private static void AtomicWrite(string path, byte[] responseBody)
+    {
+        using var response = JsonDocument.Parse(responseBody);
+        var content = JsonSerializer.SerializeToUtf8Bytes(new
+        {
+            completed_at = DateTimeOffset.UtcNow,
+            response = response.RootElement,
+        }, JsonOptions);
+        var temporary = path + ".tmp-" + Guid.NewGuid().ToString("N");
+        try
+        {
+            using var stream = new FileStream(
+                temporary, FileMode.CreateNew, FileAccess.Write, FileShare.None,
+                4096, FileOptions.WriteThrough);
+            stream.Write(content);
+            stream.Flush(flushToDisk: true);
+            File.Move(temporary, path, overwrite: false);
+        }
+        finally
+        {
+            if (File.Exists(temporary)) File.Delete(temporary);
+        }
+    }
+}

windows-node/Zcbot.WindowsNode/NodeConnectionLoop.cs

@@ -4,11 +4,20 @@ using System.Reflection;
 using System.Runtime.InteropServices;
 using System.Text;
 using System.Text.Json;
+using System.Collections.Concurrent;
 
 namespace Zcbot.WindowsNode;
 
 internal sealed class NodeConnectionLoop(NodeConfig config, Action<NodeStatus>? statusChanged = null)
 {
+    private readonly SemaphoreSlim sendLock = new(1, 1);
+    private readonly JobInboxStore jobInbox = new(NodePaths.ForCurrentMachine().JobsDirectory);
+    private readonly JobInputDownloader inputDownloader = new(
+        config, new JobInboxStore(NodePaths.ForCurrentMachine().JobsDirectory));
+    private readonly OriginWorkerRunner workerRunner = new(
+        new JobInboxStore(NodePaths.ForCurrentMachine().JobsDirectory));
+    private readonly JobOutputUploader outputUploader = new(config);
+    private readonly ConcurrentDictionary<Guid, Task> jobPipelines = new();
     private static readonly TimeSpan[] Backoff =
     [
         TimeSpan.FromSeconds(1),
@@ -49,6 +58,11 @@ internal sealed class NodeConnectionLoop(NodeConfig config, Action<NodeStatus>?
                 Report(NodeState.Offline, "服务端消息无效，等待重连");
                 Console.Error.WriteLine($"[WARN] Invalid server message: {exception.Message}");
             }
+            catch (NodeEndpointException exception)
+            {
+                Report(NodeState.Offline, "WebSocket 握手被拒绝，请检查服务端或反向代理");
+                Console.Error.WriteLine($"[WARN] WebSocket handshake rejected: {exception.Message}");
+            }
 
             var baseDelay = Backoff[Math.Min(attempt, Backoff.Length - 1)];
             attempt++;
@@ -73,17 +87,20 @@ internal sealed class NodeConnectionLoop(NodeConfig config, Action<NodeStatus>?
         {
             await socket.ConnectAsync(endpoint, cancellationToken);
         }
-        catch (WebSocketException) when (
+        catch (WebSocketException exception) when (
             socket.HttpStatusCode is HttpStatusCode.Unauthorized or HttpStatusCode.Forbidden)
         {
-            Report(NodeState.AuthenticationRequired, "身份失效，需要重新注册");
-            throw new NodeConfigurationException(
-                "Node credentials were rejected. Ask an administrator to re-register this node.");
+            throw new NodeEndpointException(
+                $"WebSocket handshake returned HTTP {(int?)socket.HttpStatusCode}. "
+                + "Verify that the server includes the current Windows Node routes and that "
+                + $"the reverse proxy forwards WebSocket Upgrade for {endpoint.AbsolutePath}. "
+                + exception.Message);
         }
         Console.WriteLine("[OK] Node connected.");
         Report(NodeState.Online, "已连接");
 
         await SendAsync(socket, "hello", RuntimePayload(), cancellationToken);
+        await ReportRecoverableJobsAsync(socket, cancellationToken);
         using var heartbeatStop = CancellationTokenSource.CreateLinkedTokenSource(cancellationToken);
         var heartbeat = HeartbeatLoopAsync(socket, heartbeatStop.Token);
         try
@@ -109,10 +126,52 @@ internal sealed class NodeConnectionLoop(NodeConfig config, Action<NodeStatus>?
         while (await timer.WaitForNextTickAsync(cancellationToken))
         {
             await SendAsync(socket, "heartbeat", RuntimePayload(), cancellationToken);
+            foreach (var job in jobInbox.ReadRecoverableJobs())
+            {
+                if (job.Terminal is JsonElement terminal
+                    && terminal.GetProperty("status").GetString() == "succeeded"
+                    && !job.UploadComplete)
+                {
+                    StartJobPipeline(socket, job);
+                }
+            }
         }
     }
 
-    private static async Task ReceiveLoopAsync(
+    private async Task ReportRecoverableJobsAsync(
+        ClientWebSocket socket, CancellationToken cancellationToken)
+    {
+        foreach (var job in jobInbox.ReadRecoverableJobs())
+        {
+            if (job.Terminal is JsonElement terminal)
+            {
+                if (terminal.GetProperty("status").GetString() == "succeeded")
+                {
+                    if (!job.UploadComplete)
+                    {
+                        StartJobPipeline(socket, job);
+                    }
+                }
+                else
+                {
+                    await SendAsync(socket, "job_terminal", terminal, cancellationToken);
+                }
+                continue;
+            }
+            await SendAsync(socket, "job_state", new
+            {
+                job_id = job.JobId,
+                lease_id = job.LeaseId,
+                request_digest = job.RequestDigest,
+                stage = "waiting_input",
+                progress = 0,
+                metrics = new { },
+            }, cancellationToken);
+            StartJobPipeline(socket, job);
+        }
+    }
+
+    private async Task ReceiveLoopAsync(
         ClientWebSocket socket, CancellationToken cancellationToken)
     {
         var buffer = new byte[16 * 1024];
@@ -122,10 +181,17 @@ internal sealed class NodeConnectionLoop(NodeConfig config, Action<NodeStatus>?
             var result = await socket.ReceiveAsync(buffer, cancellationToken);
             if (result.MessageType == WebSocketMessageType.Close)
             {
+                if ((int?)result.CloseStatus == 4003)
+                {
+                    throw new NodeConfigurationException(
+                        "节点身份已被服务端拒绝。请在管理后台确认该 Node ID 未被禁用或删除；"
+                        + "若记录不存在或身份已撤销，请清除本机身份并使用新注册码重新注册。"
+                        + $" 服务端信息：{result.CloseStatusDescription}");
+                }
                 if (result.CloseStatus == WebSocketCloseStatus.PolicyViolation)
                 {
                     throw new NodeConfigurationException(
-                        $"Node credentials or install identity were rejected: {result.CloseStatusDescription}");
+                        $"节点上报被服务端拒绝：{result.CloseStatusDescription}");
                 }
                 return;
             }
@@ -146,11 +212,222 @@ internal sealed class NodeConnectionLoop(NodeConfig config, Action<NodeStatus>?
             if (document.RootElement.TryGetProperty("type", out var type))
             {
                 Console.WriteLine($"[INFO] Server message: {type.GetString()}.");
+                if (type.GetString() == "job_offer"
+                    && document.RootElement.TryGetProperty("payload", out var payload))
+                {
+                    var offerResult = jobInbox.Accept(payload);
+                    await SendAsync(
+                        socket,
+                        offerResult.Accepted ? "job_accept" : "job_reject",
+                        offerResult.Accepted
+                            ? new
+                            {
+                                job_id = offerResult.JobId,
+                                lease_id = offerResult.LeaseId,
+                                request_digest = offerResult.RequestDigest,
+                            }
+                            : new
+                            {
+                                job_id = payload.TryGetProperty("job_id", out var jobId)
+                                    ? jobId.GetString() : "",
+                                lease_id = payload.TryGetProperty("lease_id", out var leaseId)
+                                    ? leaseId.GetString() : "",
+                                reason = offerResult.Reason,
+                            },
+                        cancellationToken);
+                    if (offerResult.Accepted)
+                    {
+                        await SendAsync(socket, "job_state", new
+                        {
+                            job_id = offerResult.JobId,
+                            lease_id = offerResult.LeaseId,
+                            request_digest = offerResult.RequestDigest,
+                            stage = "waiting_input",
+                            progress = 0,
+                            metrics = new { },
+                        }, cancellationToken);
+                        var acceptedJob = jobInbox.ReadRecoverableJobs()
+                            .Single(item => item.JobId == offerResult.JobId);
+                        StartJobPipeline(socket, acceptedJob);
+                    }
+                }
+                else if (type.GetString() == "job_cancel"
+                    && document.RootElement.TryGetProperty("payload", out var cancelPayload)
+                    && TryCancelJob(cancelPayload, out var cancelledJob))
+                {
+                    workerRunner.Cancel(cancelledJob.JobId);
+                    jobInbox.WriteTerminal(
+                        cancelledJob, "cancelled", "USER_CANCELLED", "Cancelled by user.");
+                    await SendAsync(socket, "job_terminal", new
+                    {
+                        job_id = cancelledJob.JobId,
+                        lease_id = cancelledJob.LeaseId,
+                        request_digest = cancelledJob.RequestDigest,
+                        status = "cancelled",
+                        error = new { code = "USER_CANCELLED", detail = "Cancelled by user." },
+                        artifact_manifest = Array.Empty<object>(),
+                    }, cancellationToken);
+                }
             }
             message.SetLength(0);
         }
     }
 
+    private bool TryCancelJob(JsonElement payload, out RecoverableJob job)
+    {
+        job = null!;
+        if (!payload.TryGetProperty("job_id", out var jobIdValue)
+            || !Guid.TryParse(jobIdValue.GetString(), out var jobId)
+            || !payload.TryGetProperty("lease_id", out var leaseIdValue)
+            || !Guid.TryParse(leaseIdValue.GetString(), out var leaseId)
+            || !payload.TryGetProperty("request_digest", out var digestValue))
+        {
+            return false;
+        }
+        job = jobInbox.ReadRecoverableJobs().SingleOrDefault(item =>
+            item.JobId == jobId
+            && item.LeaseId == leaseId
+            && item.RequestDigest == digestValue.GetString()
+            && item.Terminal is null)!;
+        return job is not null;
+    }
+
+    private void StartJobPipeline(ClientWebSocket socket, RecoverableJob job)
+    {
+        var completion = new TaskCompletionSource(
+            TaskCreationOptions.RunContinuationsAsynchronously);
+        if (jobPipelines.TryAdd(job.JobId, completion.Task))
+        {
+            _ = RunJobPipelineAndReleaseAsync(socket, job, completion);
+        }
+    }
+
+    private async Task RunJobPipelineAndReleaseAsync(
+        ClientWebSocket socket, RecoverableJob job, TaskCompletionSource completion)
+    {
+        try
+        {
+            await RunJobPipelineAsync(socket, job);
+        }
+        finally
+        {
+            completion.TrySetResult();
+            jobPipelines.TryRemove(job.JobId, out _);
+        }
+    }
+
+    private async Task RunJobPipelineAsync(ClientWebSocket socket, RecoverableJob job)
+    {
+        try
+        {
+            if (job.Terminal is null)
+            {
+                await inputDownloader.DownloadAsync(job, CancellationToken.None);
+                var afterDownload = jobInbox.ReadRecoverableJobs()
+                    .Single(item => item.JobId == job.JobId);
+                if (afterDownload.Terminal is JsonElement cancelledTerminal)
+                {
+                    await TrySendAsync(socket, "job_terminal", cancelledTerminal);
+                    return;
+                }
+                await TrySendAsync(socket, "job_state", new
+                {
+                    job_id = job.JobId,
+                    lease_id = job.LeaseId,
+                    request_digest = job.RequestDigest,
+                    stage = "ready_to_run",
+                    progress = 5,
+                    metrics = new { input_bytes = job.InputTransfer?.GetProperty("size_bytes").GetInt64() },
+                });
+                await TrySendAsync(socket, "job_state", new
+                {
+                    job_id = job.JobId,
+                    lease_id = job.LeaseId,
+                    request_digest = job.RequestDigest,
+                    stage = "origin_running",
+                    progress = 10,
+                    metrics = new { },
+                });
+                await workerRunner.RunAsync(job);
+            }
+            var refreshed = jobInbox.ReadRecoverableJobs()
+                .Single(item => item.JobId == job.JobId);
+            if (refreshed.Terminal is not JsonElement terminal)
+            {
+                throw new InvalidDataException("Origin worker did not create a terminal record.");
+            }
+            if (terminal.GetProperty("status").GetString() != "succeeded")
+            {
+                await TrySendAsync(socket, "job_terminal", terminal);
+                return;
+            }
+            await TrySendAsync(socket, "job_state", new
+            {
+                job_id = job.JobId,
+                lease_id = job.LeaseId,
+                request_digest = job.RequestDigest,
+                stage = "uploading_outputs",
+                progress = 90,
+                metrics = new { },
+            });
+            await outputUploader.UploadAsync(refreshed);
+        }
+        catch (Exception exception) when (
+            exception is HttpRequestException
+                or IOException
+                or JsonException
+                or UnauthorizedAccessException
+                or InvalidDataException)
+        {
+            var current = jobInbox.ReadRecoverableJobs()
+                .Single(item => item.JobId == job.JobId);
+            if (current.Terminal is JsonElement terminal
+                && terminal.GetProperty("status").GetString() == "succeeded")
+            {
+                Console.Error.WriteLine($"[WARN] Output upload deferred: {exception.Message}");
+                return;
+            }
+            jobInbox.WriteTerminal(
+                job,
+                "failed",
+                "INPUT_DOWNLOAD_FAILED",
+                exception.Message[..Math.Min(exception.Message.Length, 500)]);
+            var failedTerminal = jobInbox.ReadRecoverableJobs()
+                .Single(item => item.JobId == job.JobId).Terminal;
+            if (failedTerminal is JsonElement payload)
+            {
+                await TrySendAsync(socket, "job_terminal", payload);
+            }
+        }
+    }
+
+    private async Task TrySendTerminalAsync(ClientWebSocket socket, RecoverableJob job)
+    {
+        var terminal = jobInbox.ReadRecoverableJobs()
+            .Single(item => item.JobId == job.JobId).Terminal;
+        if (terminal is JsonElement terminalPayload)
+        {
+            await TrySendAsync(socket, "job_terminal", terminalPayload);
+        }
+    }
+
+    private async Task TrySendAsync(ClientWebSocket socket, string type, object payload)
+    {
+        if (socket.State != WebSocketState.Open)
+        {
+            return;
+        }
+        try
+        {
+            await SendAsync(socket, type, payload, CancellationToken.None);
+        }
+        catch (Exception exception) when (
+            exception is WebSocketException or IOException or ObjectDisposedException)
+        {
+            Console.Error.WriteLine($"[WARN] Job report deferred: {exception.Message}");
+        }
+    }
+
     private void Report(NodeState state, string message) =>
         statusChanged?.Invoke(NodeStatus.Create(state, message));
 
@@ -165,13 +442,22 @@ internal sealed class NodeConnectionLoop(NodeConfig config, Action<NodeStatus>?
             sent_at = DateTimeOffset.UtcNow,
             payload,
         });
-        await socket.SendAsync(
-            envelope, WebSocketMessageType.Text, endOfMessage: true, cancellationToken);
+        await sendLock.WaitAsync(cancellationToken);
+        try
+        {
+            await socket.SendAsync(
+                envelope, WebSocketMessageType.Text, endOfMessage: true, cancellationToken);
+        }
+        finally
+        {
+            sendLock.Release();
+        }
     }
 
     private object RuntimePayload()
     {
         var root = Path.GetPathRoot(Environment.SystemDirectory) ?? "C:\\";
+        var origin = OriginRuntimeProbe.Detect();
         return new
         {
             install_id = config.InstallId,
@@ -179,9 +465,17 @@ internal sealed class NodeConnectionLoop(NodeConfig config, Action<NodeStatus>?
             node_version = Assembly.GetExecutingAssembly().GetName().Version?.ToString(3) ?? "0.1.0",
             os_version = RuntimeInformation.OSDescription,
             capabilities = config.Capabilities,
-            available_slots = 1,
+            available_slots = origin.Health == "ready" && !jobInbox.HasPendingJobs ? 1 : 0,
             disk_free_bytes = new DriveInfo(root).AvailableFreeSpace,
             desktop_session = Environment.UserInteractive,
+            origin = new
+            {
+                software = origin.Software,
+                software_version = origin.SoftwareVersion,
+                adapter_version = origin.AdapterVersion,
+                health = origin.Health,
+                detail = origin.Detail,
+            },
         };
     }
 }

windows-node/Zcbot.WindowsNode/NodeModels.cs

@@ -34,15 +34,18 @@ internal sealed record EnrollResponse(
     [property: JsonPropertyName("heartbeat_seconds")] int HeartbeatSeconds,
     [property: JsonPropertyName("max_concurrency")] int MaxConcurrency);
 
-internal sealed record NodePaths(string RootDirectory, string ConfigPath)
+internal sealed record NodePaths(string RootDirectory, string ConfigPath, string JobsDirectory)
 {
     internal static NodePaths ForCurrentMachine()
     {
         var root = Path.Combine(
             Environment.GetFolderPath(Environment.SpecialFolder.CommonApplicationData),
             "Zcbot", "WindowsNode");
-        return new NodePaths(root, Path.Combine(root, "node.json"));
+        return new NodePaths(
+            root, Path.Combine(root, "node.json"), Path.Combine(root, "jobs"));
     }
 }
 
 internal sealed class NodeConfigurationException(string message) : Exception(message);
+
+internal sealed class NodeEndpointException(string message) : Exception(message);

windows-node/Zcbot.WindowsNode/OriginRuntimeProbe.cs

@@ -0,0 +1,102 @@
+using Microsoft.Win32;
+using System.Security;
+
+namespace Zcbot.WindowsNode;
+
+internal sealed record OriginRuntimeStatus(
+    string Software,
+    string? SoftwareVersion,
+    string AdapterVersion,
+    string Health,
+    string Detail);
+
+internal static class OriginRuntimeProbe
+{
+    private static readonly Lazy<OriginRuntimeStatus> Current = new(DetectCore);
+    private const string AutomationProgId = @"Origin.ApplicationSI\CLSID";
+
+    internal static OriginRuntimeStatus Detect() => Current.Value;
+
+    private static OriginRuntimeStatus DetectCore()
+    {
+        try
+        {
+            var version = FindInstalledVersion();
+            using var automationKey = Registry.ClassesRoot.OpenSubKey(AutomationProgId);
+            var automationRegistered = automationKey is not null;
+            if (version is null && !automationRegistered)
+            {
+                return Status(null, "unavailable", "未检测到 Origin/OriginPro 安装");
+            }
+            if (!automationRegistered)
+            {
+                return Status(version, "unavailable", "已检测到 Origin，但 COM 自动化组件未注册");
+            }
+            if (!Environment.UserInteractive)
+            {
+                return Status(version, "unavailable", "Origin 需要交互式 Windows 桌面会话");
+            }
+            var interpreter = OriginWorkerRuntime.ResolveInterpreter();
+            if (interpreter is null)
+            {
+                return Status(
+                    version,
+                    "unavailable",
+                    "Origin 可用，但固定 Python 运行时缺失；请配置 ZCBOT_ORIGIN_PYTHON");
+            }
+            return Status(version, "ready", $"Origin COM 与固定 Python 运行时可用（{interpreter}）");
+        }
+        catch (Exception exception) when (
+            exception is SecurityException or UnauthorizedAccessException or IOException)
+        {
+            return Status(null, "unavailable", $"Origin 运行时探测失败：{exception.Message}");
+        }
+    }
+
+    private static OriginRuntimeStatus Status(string? version, string health, string detail) =>
+        new("OriginPro", version, "0.2.0", health, detail);
+
+    private static string? FindInstalledVersion()
+    {
+        var candidates = new List<string>();
+        foreach (var hive in new[] { RegistryHive.LocalMachine, RegistryHive.CurrentUser })
+        {
+            foreach (var view in new[] { RegistryView.Registry64, RegistryView.Registry32 })
+            {
+                using var baseKey = RegistryKey.OpenBaseKey(hive, view);
+                using var uninstall = baseKey.OpenSubKey(
+                    @"SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall");
+                if (uninstall is null)
+                {
+                    continue;
+                }
+                foreach (var keyName in uninstall.GetSubKeyNames())
+                {
+                    using var product = uninstall.OpenSubKey(keyName);
+                    var name = product?.GetValue("DisplayName") as string;
+                    var publisher = product?.GetValue("Publisher") as string;
+                    if (!IsOriginProduct(name, publisher))
+                    {
+                        continue;
+                    }
+                    var version = product?.GetValue("DisplayVersion") as string;
+                    if (!string.IsNullOrWhiteSpace(version))
+                    {
+                        candidates.Add(version.Trim());
+                    }
+                }
+            }
+        }
+        return candidates.OrderByDescending(ParseVersion).ThenByDescending(x => x).FirstOrDefault();
+    }
+
+    private static bool IsOriginProduct(string? name, string? publisher) =>
+        !string.IsNullOrWhiteSpace(name)
+        && (name.Equals("Origin", StringComparison.OrdinalIgnoreCase)
+            || name.StartsWith("Origin ", StringComparison.OrdinalIgnoreCase)
+            || name.StartsWith("OriginPro", StringComparison.OrdinalIgnoreCase))
+        && (publisher?.Contains("OriginLab", StringComparison.OrdinalIgnoreCase) ?? false);
+
+    private static Version ParseVersion(string value) =>
+        Version.TryParse(value, out var version) ? version : new Version(0, 0);
+}

windows-node/Zcbot.WindowsNode/OriginWorkerRunner.cs

@@ -0,0 +1,164 @@
+using System.Collections.Concurrent;
+using System.Diagnostics;
+using System.Text;
+using System.Text.Json;
+
+namespace Zcbot.WindowsNode;
+
+internal sealed class OriginWorkerRunner(JobInboxStore inbox)
+{
+    private static readonly TimeSpan WorkerTimeout = TimeSpan.FromMinutes(30);
+    private readonly ConcurrentDictionary<Guid, Task> active = new();
+    private readonly ConcurrentDictionary<Guid, CancellationTokenSource> cancellations = new();
+
+    internal Task RunAsync(RecoverableJob job) =>
+        active.GetOrAdd(job.JobId, _ => RunOnceAsync(job, CancellationFor(job.JobId).Token));
+
+    internal void Cancel(Guid jobId)
+    {
+        CancellationFor(jobId).Cancel();
+    }
+
+    private CancellationTokenSource CancellationFor(Guid jobId) =>
+        cancellations.GetOrAdd(jobId, _ => new CancellationTokenSource());
+
+    private async Task RunOnceAsync(RecoverableJob job, CancellationToken cancellationToken)
+    {
+        try
+        {
+            var paths = NodePaths.ForCurrentMachine();
+            var jobDirectory = Path.Combine(paths.JobsDirectory, job.JobId.ToString("D"));
+            var terminalPath = Path.Combine(jobDirectory, "terminal.json");
+            if (File.Exists(terminalPath)) return;
+
+            var markerPath = Path.Combine(jobDirectory, "worker-started.json");
+            if (File.Exists(markerPath))
+            {
+                inbox.WriteTerminal(
+                    job,
+                    "failed",
+                    "NODE_RESTARTED_DURING_JOB",
+                    "The node restarted after Origin execution began and cannot prove the prior worker state.");
+                return;
+            }
+
+            var interpreter = OriginWorkerRuntime.ResolveInterpreter()
+                ?? throw new InvalidOperationException("The fixed Origin Python interpreter is unavailable.");
+            var workerScript = Path.GetFullPath(
+                Path.Combine(AppContext.BaseDirectory, "origin-worker", "worker.py"));
+            if (!File.Exists(workerScript))
+            {
+                throw new FileNotFoundException("The fixed Origin worker script is missing.", workerScript);
+            }
+            WriteMarker(markerPath, interpreter, workerScript);
+
+            var startInfo = new ProcessStartInfo
+            {
+                FileName = interpreter,
+                WorkingDirectory = jobDirectory,
+                UseShellExecute = false,
+                CreateNoWindow = true,
+                RedirectStandardOutput = true,
+                RedirectStandardError = true,
+                StandardOutputEncoding = Encoding.UTF8,
+                StandardErrorEncoding = Encoding.UTF8,
+            };
+            startInfo.ArgumentList.Add(workerScript);
+            startInfo.ArgumentList.Add(jobDirectory);
+            using var process = Process.Start(startInfo)
+                ?? throw new InvalidOperationException("The fixed Origin worker did not start.");
+            var stdout = process.StandardOutput.ReadToEndAsync();
+            var stderr = process.StandardError.ReadToEndAsync();
+            using var timeout = CancellationTokenSource.CreateLinkedTokenSource(cancellationToken);
+            timeout.CancelAfter(WorkerTimeout);
+            try
+            {
+                await process.WaitForExitAsync(timeout.Token);
+            }
+            catch (OperationCanceledException)
+            {
+                process.Kill(entireProcessTree: true);
+                if (cancellationToken.IsCancellationRequested)
+                {
+                    inbox.WriteTerminal(job, "cancelled", "USER_CANCELLED", "Cancelled by user.");
+                }
+                else
+                {
+                    inbox.WriteTerminal(job, "failed", "ORIGIN_WORKER_TIMEOUT", "Origin worker exceeded 30 minutes.");
+                }
+                return;
+            }
+            var output = await stdout;
+            var error = await stderr;
+            WriteDiagnostic(jobDirectory, output, error, process.ExitCode);
+            if (!File.Exists(terminalPath))
+            {
+                inbox.WriteTerminal(
+                    job,
+                    "failed",
+                    "ORIGIN_WORKER_NO_TERMINAL",
+                    $"Origin worker exited with code {process.ExitCode} without terminal.json.");
+            }
+        }
+        catch (Exception exception) when (
+            exception is IOException
+                or JsonException
+                or UnauthorizedAccessException
+                or InvalidOperationException)
+        {
+            inbox.WriteTerminal(job, "failed", "ORIGIN_WORKER_START_FAILED", exception.Message[..Math.Min(500, exception.Message.Length)]);
+        }
+        finally
+        {
+            active.TryRemove(job.JobId, out _);
+            if (cancellations.TryRemove(job.JobId, out var cancellation)) cancellation.Dispose();
+        }
+    }
+
+    private static void WriteMarker(string path, string interpreter, string workerScript)
+    {
+        var value = JsonSerializer.SerializeToUtf8Bytes(new
+        {
+            started_at = DateTimeOffset.UtcNow,
+            node_pid = Environment.ProcessId,
+            interpreter,
+            worker_script = workerScript,
+        });
+        using var stream = new FileStream(
+            path, FileMode.CreateNew, FileAccess.Write, FileShare.None,
+            bufferSize: 4096, FileOptions.WriteThrough);
+        stream.Write(value);
+        stream.Flush(flushToDisk: true);
+    }
+
+    private static void WriteDiagnostic(string jobDirectory, string output, string error, int exitCode)
+    {
+        var logs = Path.Combine(jobDirectory, "logs");
+        Directory.CreateDirectory(logs);
+        var value = JsonSerializer.Serialize(new
+        {
+            exit_code = exitCode,
+            stdout = output[..Math.Min(output.Length, 16 * 1024)],
+            stderr = error[..Math.Min(error.Length, 16 * 1024)],
+        });
+        File.WriteAllText(Path.Combine(logs, "worker-process.json"), value, Encoding.UTF8);
+    }
+}
+
+internal static class OriginWorkerRuntime
+{
+    internal static string? ResolveInterpreter()
+    {
+        var paths = NodePaths.ForCurrentMachine();
+        var configured = Environment.GetEnvironmentVariable("ZCBOT_ORIGIN_PYTHON");
+        var candidate = string.IsNullOrWhiteSpace(configured)
+            ? Path.Combine(paths.RootDirectory, "runtimes", "origin", "python.exe")
+            : configured;
+        if (!Path.IsPathFullyQualified(candidate)) return null;
+        var resolved = Path.GetFullPath(candidate);
+        return File.Exists(resolved)
+            && Path.GetFileName(resolved).Equals("python.exe", StringComparison.OrdinalIgnoreCase)
+            ? resolved
+            : null;
+    }
+}

windows-node/Zcbot.WindowsNode/TrayApplicationContext.cs

@@ -16,6 +16,7 @@ internal sealed class TrayApplicationContext : ApplicationContext
         this.store = store;
         form = new ConfigurationForm();
         form.RegisterRequested += RegisterAsync;
+        form.ReconnectRequested += RestartConnection;
         form.ResetIdentityRequested += ResetIdentity;
 
         statusItem = new ToolStripMenuItem("尚未注册") { Enabled = false };

windows-node/Zcbot.WindowsNode/Zcbot.WindowsNode.csproj

@@ -11,4 +11,10 @@
     <RootNamespace>Zcbot.WindowsNode</RootNamespace>
     <Version>0.1.0</Version>
   </PropertyGroup>
+  <ItemGroup>
+    <Content Include="..\origin-worker\worker.py">
+      <Link>origin-worker\worker.py</Link>
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </Content>
+  </ItemGroup>
 </Project>

windows-node/install-origin-runtime.ps1

@@ -0,0 +1,32 @@
+param(
+    [Parameter(Mandatory = $true)]
+    [string]$BootstrapPython,
+    [string]$RuntimeDirectory = "$env:ProgramData\Zcbot\WindowsNode\runtimes\origin"
+)
+
+$ErrorActionPreference = "Stop"
+$python = (Resolve-Path -LiteralPath $BootstrapPython).Path
+if ([IO.Path]::GetFileName($python) -ne "python.exe") {
+    throw "BootstrapPython must point to python.exe."
+}
+$requirements = Join-Path $PSScriptRoot "origin-worker\requirements.txt"
+if (-not (Test-Path -LiteralPath $requirements -PathType Leaf)) {
+    throw "Pinned Origin worker requirements are missing."
+}
+$runtime = [IO.Path]::GetFullPath($RuntimeDirectory)
+if ($runtime -eq [IO.Path]::GetPathRoot($runtime)) {
+    throw "RuntimeDirectory cannot be a drive root."
+}
+
+if (-not (Test-Path -LiteralPath (Join-Path $runtime "python.exe"))) {
+    & $python -m venv $runtime
+    if ($LASTEXITCODE -ne 0) { throw "Failed to create the Origin runtime." }
+}
+$runtimePython = Join-Path $runtime "python.exe"
+& $runtimePython -m pip install --requirement $requirements
+if ($LASTEXITCODE -ne 0) { throw "Failed to install the pinned Origin runtime packages." }
+& $runtimePython -c "import originpro, openpyxl; print('[OK] Origin worker Python packages are available.')"
+if ($LASTEXITCODE -ne 0) { throw "Origin runtime import verification failed." }
+
+Write-Output "[OK] Fixed Origin runtime installed: $runtimePython"
+Write-Output "[INFO] Restart zcbot Windows Node to refresh runtime health."

windows-node/origin-worker/requirements.txt

@@ -0,0 +1,2 @@
+originpro==1.1.15
+openpyxl==3.1.5

windows-node/origin-worker/worker.py

@@ -0,0 +1,252 @@
+"""Fixed Origin adapter for origin.plot@v1.
+
+This process accepts exactly one argument: a Node-created job directory. It never
+installs packages, evaluates user code, downloads data, or resolves paths from the
+request. terminal.json is its only terminal-state contract.
+"""
+from __future__ import annotations
+
+import csv
+import hashlib
+import json
+import os
+import sys
+from datetime import datetime, timezone
+from importlib.metadata import PackageNotFoundError, version
+from pathlib import Path
+from typing import Any
+
+PLOT_TYPES = {"line": "l", "scatter": "s", "line_scatter": "y"}
+FORMATS = {"opju", "png", "svg", "pdf"}
+
+
+def _atomic_json(path: Path, value: Any) -> None:
+    temporary = path.with_name(path.name + ".tmp-" + os.urandom(8).hex())
+    try:
+        with temporary.open("w", encoding="utf-8", newline="\n") as handle:
+            json.dump(value, handle, ensure_ascii=False, indent=2)
+            handle.flush()
+            os.fsync(handle.fileno())
+        os.replace(temporary, path)
+    finally:
+        temporary.unlink(missing_ok=True)
+
+
+def _read_rows(path: Path, sheet: str | None) -> tuple[list[str], list[list[Any]]]:
+    suffix = path.suffix.lower()
+    if suffix == ".csv":
+        with path.open("r", encoding="utf-8-sig", newline="") as handle:
+            rows = list(csv.reader(handle))
+        if len(rows) < 2:
+            raise ValueError("CSV_INPUT_EMPTY")
+        return [str(item) for item in rows[0]], rows[1:]
+    if suffix == ".json":
+        value = json.loads(path.read_text(encoding="utf-8"))
+        if isinstance(value, list) and value and all(isinstance(item, dict) for item in value):
+            headers = list(value[0])
+            return headers, [[item.get(name) for name in headers] for item in value]
+        if isinstance(value, dict) and value and all(isinstance(item, list) for item in value.values()):
+            headers = list(value)
+            length = max(len(value[name]) for name in headers)
+            return headers, [[value[name][index] if index < len(value[name]) else None for name in headers] for index in range(length)]
+        raise ValueError("JSON_INPUT_SHAPE_UNSUPPORTED")
+    if suffix == ".xlsx":
+        from openpyxl import load_workbook
+
+        workbook = load_workbook(path, read_only=True, data_only=True)
+        try:
+            worksheet = workbook[sheet] if sheet else workbook.active
+            rows = list(worksheet.iter_rows(values_only=True))
+        finally:
+            workbook.close()
+        if len(rows) < 2:
+            raise ValueError("XLSX_INPUT_EMPTY")
+        return [str(item or "") for item in rows[0]], [list(row) for row in rows[1:]]
+    raise ValueError("INPUT_TYPE_UNSUPPORTED")
+
+
+def _column_index(headers: list[str], value: Any, field: str) -> int:
+    if not isinstance(value, str) or value not in headers:
+        raise ValueError(f"{field.upper()}_COLUMN_NOT_FOUND")
+    return headers.index(value)
+
+
+def _manifest(path: Path, media_type: str) -> dict[str, Any]:
+    return {
+        "artifact_id": {
+            "project.opju": "project",
+            "figure.png": "figure_png",
+            "figure.svg": "figure_svg",
+            "figure.pdf": "figure_pdf",
+            "plot-spec.json": "plot_spec",
+            "provenance.json": "provenance",
+        }[path.name],
+        "filename": path.name,
+        "media_type": media_type,
+        "size_bytes": path.stat().st_size,
+        "sha256": _file_sha256(path),
+    }
+
+
+def _file_sha256(path: Path) -> str:
+    digest = hashlib.sha256()
+    with path.open("rb") as handle:
+        for chunk in iter(lambda: handle.read(1024 * 1024), b""):
+            digest.update(chunk)
+    return digest.hexdigest()
+
+
+def _axis_title(axis: Any, fallback: str) -> str:
+    if not isinstance(axis, dict):
+        return fallback
+    title = str(axis.get("title") or fallback)
+    unit = str(axis.get("unit") or "")
+    return f"{title} ({unit})" if unit else title
+
+
+def _validate_artifact(path: Path, extension: str) -> None:
+    if not path.is_file() or path.stat().st_size == 0:
+        raise RuntimeError(f"{extension.upper()}_EXPORT_EMPTY")
+    head = path.read_bytes()[:1024]
+    if extension == "png" and not head.startswith(b"\x89PNG\r\n\x1a\n"):
+        raise RuntimeError("PNG_EXPORT_INVALID")
+    if extension == "pdf" and not head.startswith(b"%PDF-"):
+        raise RuntimeError("PDF_EXPORT_INVALID")
+    if extension == "svg" and b"<svg" not in head.lower():
+        raise RuntimeError("SVG_EXPORT_INVALID")
+    if extension == "opju" and len(head) < 64:
+        raise RuntimeError("OPJU_EXPORT_INVALID")
+
+
+def run(job_dir: Path) -> list[dict[str, Any]]:
+    job_dir = job_dir.resolve(strict=True)
+    request_record = json.loads((job_dir / "request" / "request.json").read_text(encoding="utf-8"))
+    request = request_record["request"]
+    input_files = [path for path in (job_dir / "input").iterdir() if path.is_file() and not path.name.startswith(".")]
+    if len(input_files) != 1:
+        raise ValueError("INPUT_FILE_COUNT_INVALID")
+    headers, rows = _read_rows(input_files[0], request["input"].get("sheet"))
+    plot_spec = request["plot"]
+    plot_type = plot_spec["type"]
+    if plot_type not in PLOT_TYPES:
+        raise ValueError("PLOT_TYPE_NOT_IMPLEMENTED")
+    x_index = _column_index(headers, plot_spec.get("x"), "x")
+    y_names = plot_spec.get("y")
+    if isinstance(y_names, str):
+        y_names = [y_names]
+    if not isinstance(y_names, list) or not y_names:
+        raise ValueError("Y_COLUMNS_REQUIRED")
+    y_indexes = [_column_index(headers, name, "y") for name in y_names]
+
+    import originpro as op
+
+    output = job_dir / "output"
+    output.mkdir(exist_ok=True)
+    op.set_show(False)
+    try:
+        op.new()
+        worksheet = op.new_sheet("w", lname="Data")
+        for index, header in enumerate(headers):
+            worksheet.from_list(index, [row[index] if index < len(row) else None for row in rows], lname=header)
+        graph = op.new_graph(template={"line": "line", "scatter": "scatter", "line_scatter": "linesymb"}[plot_type])
+        layer = graph[0]
+        for y_index in y_indexes:
+            layer.add_plot(worksheet, coly=y_index, colx=x_index, type=PLOT_TYPES[plot_type])
+        layer.rescale()
+        layer.axis("x").title = _axis_title(plot_spec.get("x_axis"), str(plot_spec.get("x") or "X"))
+        layer.axis("y").title = _axis_title(plot_spec.get("y_axis"), "Y")
+        if plot_spec.get("title"):
+            title = layer.add_label(str(plot_spec["title"]))
+            title.set_int("fsize", 18)
+            title.set_int("left", 2200)
+            title.set_int("top", 120)
+        formats = request["output"]["formats"]
+        if any(item not in FORMATS for item in formats):
+            raise ValueError("OUTPUT_FORMAT_UNSUPPORTED")
+        artifacts: list[dict[str, Any]] = []
+        if "opju" in formats:
+            project = output / "project.opju"
+            op.save(str(project))
+            _validate_artifact(project, "opju")
+            artifacts.append(_manifest(project, "application/x-origin-project"))
+        media = {"png": "image/png", "svg": "image/svg+xml", "pdf": "application/pdf"}
+        dpi = request["output"].get("dpi", 300)
+        if not isinstance(dpi, int) or isinstance(dpi, bool) or not 72 <= dpi <= 1200:
+            raise ValueError("OUTPUT_DPI_INVALID")
+        pixel_width = round(dpi * 160 / 25.4)
+        for extension in ("png", "svg", "pdf"):
+            if extension in formats:
+                target = output / f"figure.{extension}"
+                exported = Path(graph.save_fig(
+                    str(target),
+                    type=extension,
+                    width=pixel_width if extension == "png" else 0,
+                    ratio=100 if extension in {"svg", "pdf"} else 0,
+                )).resolve()
+                if exported != target.resolve() or not target.is_file():
+                    raise RuntimeError(f"{extension.upper()}_EXPORT_FAILED")
+                _validate_artifact(target, extension)
+                artifacts.append(_manifest(target, media[extension]))
+        try:
+            originpro_version = version("originpro")
+        except PackageNotFoundError:
+            originpro_version = "embedded"
+        provenance = {
+            "adapter_version": "0.2.0",
+            "originpro_version": originpro_version,
+            "request_digest": request_record["request_digest"],
+            "input_sha256": _file_sha256(input_files[0]),
+            "requested_dpi": dpi,
+            "png_pixel_width": pixel_width,
+        }
+        plot_spec_path = output / "plot-spec.json"
+        provenance_path = output / "provenance.json"
+        _atomic_json(plot_spec_path, request)
+        _atomic_json(provenance_path, provenance)
+        artifacts.append(_manifest(plot_spec_path, "application/json"))
+        artifacts.append(_manifest(provenance_path, "application/json"))
+        return artifacts
+    finally:
+        if op.oext:
+            op.exit()
+
+
+def main() -> int:
+    if len(sys.argv) != 2:
+        print("[ERR] Usage: worker.py <job-directory>", file=sys.stderr)
+        return 2
+    job_dir = Path(sys.argv[1])
+    request_record: dict[str, Any] = {}
+    try:
+        request_record = json.loads((job_dir / "request" / "request.json").read_text(encoding="utf-8"))
+        artifacts = run(job_dir)
+        terminal = {
+            "job_id": request_record["job_id"],
+            "lease_id": request_record["lease_id"],
+            "request_digest": request_record["request_digest"],
+            "status": "succeeded",
+            "error": {},
+            "artifact_manifest": artifacts,
+            "terminal_at": datetime.now(timezone.utc).isoformat(),
+        }
+        _atomic_json(job_dir / "artifacts.json", artifacts)
+        _atomic_json(job_dir / "terminal.json", terminal)
+        print("[OK] Origin job completed.")
+        return 0
+    except Exception as exception:
+        terminal = {
+            "job_id": request_record.get("job_id", ""),
+            "lease_id": request_record.get("lease_id", ""),
+            "request_digest": request_record.get("request_digest", ""),
+            "status": "failed",
+            "error": {"code": type(exception).__name__, "detail": str(exception)[:500]},
+            "artifact_manifest": [],
+            "terminal_at": datetime.now(timezone.utc).isoformat(),
+        }
+        _atomic_json(job_dir / "terminal.json", terminal)
+        print(f"[ERR] {type(exception).__name__}: {exception}", file=sys.stderr)
+        return 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())