helloJamest · helloJamest · Jul 5, 2026 · Jul 5, 2026
diff --git a/.agents/skills/analyze-issue/SKILL.md b/.agents/skills/analyze-issue/SKILL.md
@@ -58,7 +58,7 @@ gh issue view <issue_number> --repo helloJamest/FinAgent --comments
 
 ### Step 5: 生成分析文档
 
-保存到 `.Codex/reviews/issues/issue-<number>.md`
+保存到 `.claude/reviews/issues/issue-<number>.md`
 
 ## Output Document Format
 

diff --git a/.agents/skills/analyze-pr/SKILL.md b/.agents/skills/analyze-pr/SKILL.md
@@ -64,7 +64,7 @@ gh run view <run_id> --log-failed
 
 ### Step 5: 生成评审文档
 
-保存到 `.Codex/reviews/prs/pr-<number>.md`
+保存到 `.claude/reviews/prs/pr-<number>.md`
 
 ## Output Document Format
 

diff --git a/.agents/skills/fix-issue/SKILL.md b/.agents/skills/fix-issue/SKILL.md
@@ -18,7 +18,7 @@
 
 ### Step 1: 确认分析基线
 
-检查 `.Codex/reviews/issues/issue-<number>.md` 是否存在；如果不存在，先补做 issue 分析或在本次修复中补齐最小分析结论。
+检查 `.claude/reviews/issues/issue-<number>.md` 是否存在；如果不存在，先补做 issue 分析或在本次修复中补齐最小分析结论。
 
 ### Step 2: 选择安全的工作方式
 
@@ -50,7 +50,7 @@
 
 ### Step 5: 更新 issue 分析文档
 
-在 `.Codex/reviews/issues/issue-<number>.md` 中补充：
+在 `.claude/reviews/issues/issue-<number>.md` 中补充：
 
 ```markdown
 ## Fix Implementation

diff --git a/.github/PULL_REQUEST_TEMPLATE.md b/.github/PULL_REQUEST_TEMPLATE.md
@@ -0,0 +1,32 @@
+## PR Type
+
+- [ ] 新功能
+- [ ] 改进
+- [ ] 修复
+- [ ] 文档
+- [ ] 测试
+- [ ] chore
+
+## Background And Problem
+
+<!-- What problem does this PR solve? Link related context when possible. -->
+
+## Scope Of Change
+
+<!-- List the files/modules touched and keep unrelated changes out. -->
+
+## Issue Link
+
+<!-- e.g. Fixes #123, or "None". -->
+
+## Verification Commands And Results
+
+<!-- Paste commands and concise results. If skipped, explain why. -->
+
+## Compatibility And Risk
+
+<!-- API/schema/config/deployment/user-visible compatibility notes. -->
+
+## Rollback Plan
+
+<!-- How to revert or disable this safely. -->
diff --git a/.github/copilot-instructions.md b/.github/copilot-instructions.md
@@ -0,0 +1,8 @@
+# Repository Instructions For Copilot And Coding Agents
+
+Canonical source: `AGENTS.md`.
+
+`CLAUDE.md` is a compatibility entry that must point to `AGENTS.md`.
+Repository skills live in `.claude/skills/` and are mirrored in `.agents/skills/`.
+
+If any instruction here conflicts with `AGENTS.md`, follow `AGENTS.md`.
diff --git a/.github/instructions/backend.instructions.md b/.github/instructions/backend.instructions.md
@@ -0,0 +1,9 @@
+# Backend Instructions
+
+Canonical source: `AGENTS.md`.
+
+- Put backend changes in `src/`, `data_provider/`, `api/`, or `bot/`.
+- Prefer existing services, repositories, schemas, and scripts.
+- Validate Python backend changes with `bash scripts/ci_gate.sh` when practical.
+- For small Python edits, at minimum run `python -m py_compile <changed_python_files>`.
+- Keep data-source fallback and notification failure paths fail-open unless the task explicitly says otherwise.
diff --git a/.github/instructions/client.instructions.md b/.github/instructions/client.instructions.md
@@ -0,0 +1,9 @@
+# Client Instructions
+
+Canonical source: `AGENTS.md`.
+
+- Put Web changes in `apps/finagent-web/`.
+- Put desktop changes in `apps/finagent-desktop/`.
+- For Web changes, run `npm ci`, `npm run lint`, and `npm run build` from `apps/finagent-web/`.
+- For desktop changes, build the Web app first, then run the desktop build.
+- Keep API/schema changes backward compatible where possible.
diff --git a/.github/instructions/governance.instructions.md b/.github/instructions/governance.instructions.md
@@ -0,0 +1,9 @@
+# Governance Instructions
+
+Canonical source: `AGENTS.md`.
+
+- Do not commit, tag, or push without explicit user confirmation.
+- Keep `.env.example` and relevant docs in sync with new configuration.
+- Update `docs/CHANGELOG.md` for user-visible, CLI/API, deployment, workflow, notification, or report changes.
+- Keep `[Unreleased]` flat: `- [类型] 描述`, with type one of `新功能`, `改进`, `修复`, `文档`, `测试`, or `chore`.
+- Run `python scripts/check_ai_assets.py` after changing AI collaboration assets.
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -0,0 +1,87 @@
+name: CI
+
+on:
+  pull_request:
+  push:
+    branches:
+      - main
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+concurrency:
+  group: ci-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  ai-governance:
+    name: AI asset governance
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Check AI collaboration assets
+        run: python scripts/check_ai_assets.py
+
+  backend-gate:
+    name: Backend gate
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+          cache: pip
+          cache-dependency-path: |
+            requirements.txt
+            requirements-ci.txt
+
+      - name: Install backend CI dependencies
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install -r requirements-ci.txt
+
+      - name: Run backend gate
+        run: bash scripts/ci_gate.sh
+
+  docker-build:
+    name: Docker build
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Build Docker image
+        run: docker build -f docker/Dockerfile -t finagent-ci .
+
+      - name: Smoke import key modules
+        run: docker run --rm finagent-ci python -c "import api.app; import src.core.pipeline; import data_provider.base"
+
+  web-gate:
+    name: Web gate
+    runs-on: ubuntu-latest
+    defaults:
+      run:
+        working-directory: apps/finagent-web
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-node@v4
+        with:
+          node-version: "20"
+          cache: npm
+          cache-dependency-path: apps/finagent-web/package-lock.json
+
+      - name: Install web dependencies
+        run: npm ci
+
+      - name: Lint web
+        run: npm run lint
+
+      - name: Build web
+        run: npm run build
diff --git a/.github/workflows/network-smoke.yml b/.github/workflows/network-smoke.yml
@@ -0,0 +1,37 @@
+name: Network Smoke
+
+on:
+  schedule:
+    - cron: "0 2 * * 1"
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  network-smoke:
+    name: Network smoke
+    runs-on: ubuntu-latest
+    continue-on-error: true
+    timeout-minutes: 60
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+          cache: pip
+          cache-dependency-path: |
+            requirements.txt
+            requirements-ci.txt
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install -r requirements-ci.txt
+
+      - name: Run network-marked pytest
+        run: python -m pytest -m network
+
+      - name: Run quick deterministic smoke
+        run: bash test.sh quick --dry-run --no-notify
diff --git a/.github/workflows/pr-review.yml b/.github/workflows/pr-review.yml
@@ -0,0 +1,38 @@
+name: PR Review
+
+on:
+  pull_request:
+    types: [opened, edited, synchronize, reopened]
+  workflow_dispatch:
+
+permissions:
+  contents: read
+  pull-requests: read
+
+jobs:
+  pr-static-checks:
+    name: PR static checks
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Check AI assets
+        run: python scripts/check_ai_assets.py
+
+      - name: Check PR template exists
+        run: test -f .github/PULL_REQUEST_TEMPLATE.md
+
+  ai-review-hook:
+    name: Optional AI review hook
+    runs-on: ubuntu-latest
+    continue-on-error: ${{ vars.AI_REVIEW_STRICT != 'true' }}
+    if: ${{ vars.AI_REVIEW_ENABLED == 'true' }}
+    steps:
+      - name: No reviewer configured
+        run: |
+          echo "AI_REVIEW_ENABLED is true, but no repository AI reviewer command is configured."
+          echo "Keep this job non-blocking unless AI_REVIEW_STRICT=true is intentionally enabled."
diff --git a/AGENTS.md b/AGENTS.md
@@ -29,9 +29,9 @@
 - `AGENTS.md` 是仓库内 AI 协作规则的唯一真源。
 - `CLAUDE.md` 必须是指向 `AGENTS.md` 的软链接，用于兼容 Claude 生态。
 - `.github/copilot-instructions.md` 与 `.github/instructions/*.instructions.md` 是 GitHub Copilot / Coding Agent 的镜像或分层补充；若与本文件冲突，以 `AGENTS.md` 为准。
-- 仓库协作 skill 存放在 `.claude/skills/`，分析产物存放在 `.claude/reviews/`；前者可以入库，后者默认视为本地产物。
+- 仓库协作 skill 存放在 `.claude/skills/`，`.agents/skills/` 是同义镜像；分析产物存放在 `.claude/reviews/`，默认视为本地产物。
 - 根目录 `SKILL.md` 与 `docs/openclaw-skill-integration.md` 属于产品或外部集成说明，不是仓库协作规则真源。
-- 若未来新增 `.agents/skills/` 或其他 agent 专用目录，必须先明确单一真源，再通过脚本或镜像同步；禁止手工长期维护多份同义内容。
+- 若未来新增其他 agent 专用目录，必须先明确单一真源，再通过脚本或镜像同步；禁止手工长期维护多份同义内容。
 - 修改 AI 协作治理资产时，执行：
 
 ```bash

diff --git a/CLAUDE.md b/CLAUDE.md
diff --git a/CLAUDE.md b/CLAUDE.md
@@ -0,0 +1 @@
+AGENTS.md
diff --git a/docs/CHANGELOG.md b/docs/CHANGELOG.md
@@ -14,7 +14,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/).
 - [修复] 补回 GitHub Actions 每日股票分析工作流，恢复 fork 后手动/定时运行能力（fixes #31）。
 - [新功能] 打板策略（screen_board_play）：通过 AKShare 获取龙虎榜、涨停池、连板天梯、涨停概念数据，自动筛选 1进2/2进3/3进4 晋级股，输出板块策略 + 连板预测 + 个股买入建议
 - [改进] 打板策略结果在 UI 中按连板层级（1进2 / 2进3 / 3进4）分组展示，不同层级使用不同颜色区分
-- [新增] API 客户端新增 `getBoardData` 方法支持直接获取打板策略原始数据
+- [新功能] API 客户端新增 `getBoardData` 方法支持直接获取打板策略原始数据
+- [chore] 恢复 CI、网络冒烟与 PR 静态检查工作流，并补齐 GitHub Copilot / Coding Agent 协作资产镜像
+- [测试] 新增 AI 资产、CI 工作流与 Changelog `[Unreleased]` 格式守卫，防止协作治理漂移
 - [修复] `AGENT_MAX_STEPS` 在 orchestrator 多 Agent 模式下改为作为各子 Agent 的步数上限而非硬覆盖；TechnicalAgent 等高默认值 Agent 会被封顶，低默认值 Agent 保持原值，减少不必要的 LLM 调用膨胀与配额消耗。
 - [修复] **MiniMax-M2.7 模型连接测试支持** — 修复 LLM 通道连接测试在 MiniMax-M2.7 模型下返回 "Empty response" 的问题；增加了 `max_tokens` 上限（8→256）以容纳 MiniMax 思考过程，并添加 `content_blocks` 格式解析逻辑统一处理 MiniMax 响应格式差异。
 - [修复] 移除 `HistoryItem` 与 `ReportSummary` 响应 Schema 中 `sentiment_score` 的 `ge=0/le=100` 约束（fixes #942）——历史库中存储的超范围负值或大于 100 的情绪评分不再触发 Pydantic ValidationError，历史列表与详情接口恢复正常返回。

diff --git a/docs/CONTRIBUTING.md b/docs/CONTRIBUTING.md
@@ -78,10 +78,12 @@ docs: 更新 README 部署说明
 
 | 检查项 | 说明 | 必须通过 |
 |--------|------|:--------:|
+| ai-governance | 校验 `AGENTS.md` / `CLAUDE.md` / `.github` 指令 / repository skills 关系 | ✅ |
 | backend-gate | `scripts/ci_gate.sh`（py_compile + flake8 严重错误 + 本地核心脚本 + offline pytest） | ✅ |
 | docker-build | Docker 镜像构建与关键模块导入 smoke | ✅ |
 | web-gate | 前端变更时执行 `npm run lint` + `npm run build` | ✅（触发时） |
 | network-smoke | 定时/手动执行 `pytest -m network` + `test.sh quick`（非阻断） | ❌（观测项） |
+| pr-review | PR 模板与协作资产静态检查，AI review hook 可选启用 | ❌（辅助项） |
 
 **本地运行检查：**
 

diff --git a/docs/CONTRIBUTING_EN.md b/docs/CONTRIBUTING_EN.md
@@ -80,9 +80,11 @@ After opening a PR, CI will automatically run the following PR checks:
 
 | Check | Description | Required |
 |-------|-------------|:--------:|
+| `ai-governance` | Checks `AGENTS.md` / `CLAUDE.md` / `.github` instructions / repository skills alignment | ✅ |
 | `backend-gate` | `scripts/ci_gate.sh` — py_compile + flake8 critical errors + `./test.sh code` + `./test.sh yfinance` + offline pytest | ✅ |
 | `docker-build` | Docker image build and key module import smoke test | ✅ |
 | `web-gate` | `npm run lint` + `npm run build` (triggered when `apps/finagent-web/` changes) | ✅ (when triggered) |
+| `pr-review` | PR template and collaboration asset static checks; optional AI review hook | ❌ (auxiliary) |
 
 Separately, the repository also has a non-blocking `network-smoke` workflow in `.github/workflows/network-smoke.yml`, but it is only triggered by `schedule` and `workflow_dispatch`, not by pull requests.