diff --git a/.gitea/workflows/issue-branch-preview.yml b/.gitea/workflows/issue-branch-preview.yml
new file mode 100644
index 0000000..4a1e78a
--- /dev/null
+++ b/.gitea/workflows/issue-branch-preview.yml
@@ -0,0 +1,164 @@
+name: issue-branch-preview
+
+on:
+  push:
+    branches-ignore:
+      - main
+  workflow_dispatch:
+    inputs:
+      branch:
+        description: "Target branch (optional, default current ref)"
+        required: false
+        type: string
+      issue:
+        description: "Issue number (optional, auto-parse from branch)"
+        required: false
+        type: string
+
+jobs:
+  allocate-and-deploy:
+    runs-on: ubuntu-latest
+    env:
+      PREVIEW_SLOTS: ${{ vars.PREVIEW_SLOTS }}
+      PREVIEW_URL_TEMPLATE: ${{ vars.PREVIEW_URL_TEMPLATE }}
+      PREVIEW_TTL_HOURS: ${{ vars.PREVIEW_TTL_HOURS }}
+      PREVIEW_STATE_FILE: .tmp/preview-slots.json
+      CLIENT_DEPLOY_CMD: ${{ vars.CLIENT_DEPLOY_CMD }}
+      SERVER_DEPLOY_CMD: ${{ vars.SERVER_DEPLOY_CMD }}
+      FULL_STACK_DEPLOY_CMD: ${{ vars.FULL_STACK_DEPLOY_CMD }}
+      INFRA_APPLY_CMD: ${{ vars.INFRA_APPLY_CMD }}
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Setup Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Resolve branch and issue
+        id: target
+        shell: bash
+        run: |
+          BRANCH="${{ inputs.branch }}"
+          ISSUE_INPUT="${{ inputs.issue }}"
+          if [ -z "$BRANCH" ]; then
+            BRANCH="${GITHUB_REF_NAME:-$(git rev-parse --abbrev-ref HEAD)}"
+          fi
+
+          ISSUE_ID="$ISSUE_INPUT"
+          if [ -z "$ISSUE_ID" ]; then
+            ISSUE_ID="$(echo "$BRANCH" | sed -nE 's#^issue[-/ ]?([0-9]+).*$#\1#p')"
+          fi
+          if [ -z "$ISSUE_ID" ]; then
+            ISSUE_ID="$(echo "$BRANCH" | sed -nE 's#^.*/([0-9]+).*$#\1#p')"
+          fi
+          if [ -z "$ISSUE_ID" ]; then
+            ISSUE_ID="0"
+          fi
+
+          echo "branch=$BRANCH" >> "$GITHUB_OUTPUT"
+          echo "issue=$ISSUE_ID" >> "$GITHUB_OUTPUT"
+
+      - name: Detect change scope
+        id: scope
+        shell: bash
+        run: |
+          git fetch origin main --depth=1 || true
+          mkdir -p .tmp
+          python skills/gitea-issue-devops-agent/scripts/change_scope.py \
+            --repo-path . \
+            --base-ref origin/main \
+            --head-ref "${{ steps.target.outputs.branch }}" > .tmp/change-scope.json
+          SCOPE="$(python -c "import json;print(json.load(open('.tmp/change-scope.json', encoding='utf-8'))['scope'])")"
+          echo "scope=$SCOPE" >> "$GITHUB_OUTPUT"
+          cat .tmp/change-scope.json
+
+      - name: Allocate preview slot
+        id: slot
+        shell: bash
+        run: |
+          SLOTS="${PREVIEW_SLOTS:-preview-a,preview-b}"
+          TTL="${PREVIEW_TTL_HOURS:-24}"
+          URL_TEMPLATE="${PREVIEW_URL_TEMPLATE:-https://{slot}.qa.example.com}"
+          mkdir -p .tmp
+
+          python skills/gitea-issue-devops-agent/scripts/preview_slot_allocator.py \
+            --state-file "$PREVIEW_STATE_FILE" \
+            --slots "$SLOTS" \
+            --repo "${GITHUB_REPOSITORY}" \
+            --issue "${{ steps.target.outputs.issue }}" \
+            --branch "${{ steps.target.outputs.branch }}" \
+            --ttl-hours "$TTL" \
+            --url-template "$URL_TEMPLATE" \
+            --evict-oldest > .tmp/slot-allocation.json
+
+          SLOT="$(python -c "import json;d=json.load(open('.tmp/slot-allocation.json', encoding='utf-8'));print(d.get('allocation',{}).get('slot',''))")"
+          URL="$(python -c "import json;d=json.load(open('.tmp/slot-allocation.json', encoding='utf-8'));print(d.get('allocation',{}).get('url',''))")"
+          echo "slot=$SLOT" >> "$GITHUB_OUTPUT"
+          echo "url=$URL" >> "$GITHUB_OUTPUT"
+          cat .tmp/slot-allocation.json
+
+      - name: Deploy by scope
+        shell: bash
+        run: |
+          set -euo pipefail
+          SCOPE="${{ steps.scope.outputs.scope }}"
+          run_or_echo () {
+            local cmd="$1"
+            local fallback="$2"
+            if [ -n "$cmd" ]; then
+              bash -lc "$cmd"
+            else
+              echo "$fallback"
+            fi
+          }
+
+          case "$SCOPE" in
+            skip)
+              echo "Scope=skip: docs/tests-only or no changes, deployment skipped."
+              ;;
+            client_only)
+              run_or_echo "${CLIENT_DEPLOY_CMD:-}" "Scope=client_only: set repo var CLIENT_DEPLOY_CMD."
+              ;;
+            server_only)
+              run_or_echo "${SERVER_DEPLOY_CMD:-}" "Scope=server_only: set repo var SERVER_DEPLOY_CMD."
+              ;;
+            full_stack)
+              run_or_echo "${FULL_STACK_DEPLOY_CMD:-}" "Scope=full_stack: set repo var FULL_STACK_DEPLOY_CMD."
+              ;;
+            infra_only)
+              run_or_echo "${INFRA_APPLY_CMD:-}" "Scope=infra_only: set repo var INFRA_APPLY_CMD."
+              ;;
+            *)
+              echo "Unknown scope: $SCOPE"
+              exit 1
+              ;;
+          esac
+
+      - name: Persist preview slot state
+        shell: bash
+        run: |
+          if [ ! -f "$PREVIEW_STATE_FILE" ]; then
+            exit 0
+          fi
+          if [ -z "$(git status --porcelain -- "$PREVIEW_STATE_FILE")" ]; then
+            exit 0
+          fi
+
+          git config user.name "gitea-actions"
+          git config user.email "gitea-actions@local"
+          git add "$PREVIEW_STATE_FILE"
+          git commit -m "chore: update preview slot state [skip ci]" || true
+          git push || true
+
+      - name: Summary
+        shell: bash
+        run: |
+          echo "branch: ${{ steps.target.outputs.branch }}"
+          echo "issue: ${{ steps.target.outputs.issue }}"
+          echo "scope: ${{ steps.scope.outputs.scope }}"
+          echo "slot: ${{ steps.slot.outputs.slot }}"
+          echo "url: ${{ steps.slot.outputs.url }}"
diff --git a/.gitea/workflows/preview-slot-reclaim.yml b/.gitea/workflows/preview-slot-reclaim.yml
new file mode 100644
index 0000000..5db4d4a
--- /dev/null
+++ b/.gitea/workflows/preview-slot-reclaim.yml
@@ -0,0 +1,83 @@
+name: preview-slot-reclaim
+
+on:
+  schedule:
+    - cron: "15 * * * *"
+  workflow_dispatch:
+    inputs:
+      repo:
+        description: "owner/repo (optional)"
+        required: false
+        type: string
+      branch:
+        description: "Branch to release (optional)"
+        required: false
+        type: string
+      issue:
+        description: "Issue number to release (optional)"
+        required: false
+        type: string
+
+jobs:
+  reclaim:
+    runs-on: ubuntu-latest
+    env:
+      PREVIEW_SLOTS: ${{ vars.PREVIEW_SLOTS }}
+      PREVIEW_STATE_FILE: .tmp/preview-slots.json
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Setup Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Release target or prune expired
+        shell: bash
+        run: |
+          SLOTS="${PREVIEW_SLOTS:-preview-a,preview-b}"
+          REPO_INPUT="${{ inputs.repo }}"
+          BRANCH_INPUT="${{ inputs.branch }}"
+          ISSUE_INPUT="${{ inputs.issue }}"
+
+          mkdir -p .tmp
+
+          if [ -n "$REPO_INPUT" ] || [ -n "$BRANCH_INPUT" ] || [ -n "$ISSUE_INPUT" ]; then
+            ISSUE_FLAG=""
+            if [ -n "$ISSUE_INPUT" ]; then
+              ISSUE_FLAG="--issue $ISSUE_INPUT"
+            fi
+            python skills/gitea-issue-devops-agent/scripts/preview_slot_allocator.py \
+              --state-file "$PREVIEW_STATE_FILE" \
+              --slots "$SLOTS" \
+              --repo "$REPO_INPUT" \
+              --branch "$BRANCH_INPUT" \
+              $ISSUE_FLAG \
+              --release > .tmp/slot-reclaim.json
+          else
+            python skills/gitea-issue-devops-agent/scripts/preview_slot_allocator.py \
+              --state-file "$PREVIEW_STATE_FILE" \
+              --slots "$SLOTS" \
+              --list > .tmp/slot-reclaim.json
+          fi
+
+          cat .tmp/slot-reclaim.json
+
+      - name: Persist state
+        shell: bash
+        run: |
+          if [ ! -f "$PREVIEW_STATE_FILE" ]; then
+            exit 0
+          fi
+          if [ -z "$(git status --porcelain -- "$PREVIEW_STATE_FILE")" ]; then
+            exit 0
+          fi
+
+          git config user.name "gitea-actions"
+          git config user.email "gitea-actions@local"
+          git add "$PREVIEW_STATE_FILE"
+          git commit -m "chore: reclaim preview slots [skip ci]" || true
+          git push || true
diff --git a/README.md b/README.md
index 1b406f0..42ad0cd 100644
--- a/README.md
+++ b/README.md
@@ -1,3 +1,55 @@
-# devops-skills
+# DevOps Skills
 
-Devops驱动开发场景投喂工具
\ No newline at end of file
+面向 **Gitea Issue 驱动交付** 的技能仓库，内置 `gitea-issue-devops-agent`，支持：
+
+- 根据 issue 指定分支执行修复与提测
+- 分支级预览环境槽位分配与回收
+- 按变更范围智能部署（避免无意义重启服务端）
+- 自动 / 半自动 / 全人工 三种协作模式
+- 图片类 issue 证据抓取与审阅
+
+## 文档网页
+
+- 站点文件：`site/index.html`
+- 仓库内查看：`https://fun-md.com/Fun_MD/devops-skills/src/branch/main/site/index.html`
+- 原始页面：`https://fun-md.com/Fun_MD/devops-skills/raw/branch/main/site/index.html`
+
+## 技能路径
+
+- `skills/gitea-issue-devops-agent/SKILL.md`
+
+## 安装
+
+```bash
+git clone https://fun-md.com/Fun_MD/devops-skills.git
+cd devops-skills
+mkdir -p ~/.codex/skills
+cp -r skills/gitea-issue-devops-agent ~/.codex/skills/
+```
+
+Windows PowerShell:
+
+```powershell
+git clone https://fun-md.com/Fun_MD/devops-skills.git
+cd devops-skills
+New-Item -ItemType Directory -Force $HOME\.codex\skills | Out-Null
+Copy-Item .\skills\gitea-issue-devops-agent $HOME\.codex\skills\gitea-issue-devops-agent -Recurse -Force
+```
+
+## 核心脚本
+
+- `skills/gitea-issue-devops-agent/scripts/issue_audit.py`
+  - issue 拉取、质量评分、去重、附件/图片抓取
+- `skills/gitea-issue-devops-agent/scripts/change_scope.py`
+  - 识别 `skip/client_only/server_only/full_stack/infra_only`
+- `skills/gitea-issue-devops-agent/scripts/preview_slot_allocator.py`
+  - 分支预览槽位分配、复用、释放、TTL 回收
+
+## 工作流模板
+
+仓库提供 `.gitea/workflows` 示例，可直接接入：
+
+- `.gitea/workflows/issue-branch-preview.yml`
+- `.gitea/workflows/preview-slot-reclaim.yml`
+
+用于实现“分配槽位 + 按变更范围部署 + 自动回收”。
diff --git a/site/index.html b/site/index.html
new file mode 100644
index 0000000..f3f543f
--- /dev/null
+++ b/site/index.html
@@ -0,0 +1,204 @@
+<!doctype html>
+<html lang="zh-CN">
+<head>
+  <meta charset="utf-8" />
+  <meta name="viewport" content="width=device-width, initial-scale=1" />
+  <title>Gitea Issue DevOps Agent</title>
+  <style>
+    :root {
+      --bg: #f4f7fb;
+      --panel: #ffffff;
+      --ink: #0f172a;
+      --muted: #475569;
+      --brand: #0ea5e9;
+      --brand-2: #14b8a6;
+      --line: #dbe5ef;
+      --ok: #16a34a;
+    }
+    * { box-sizing: border-box; }
+    body {
+      margin: 0;
+      font-family: "Segoe UI", "PingFang SC", "Microsoft YaHei", sans-serif;
+      color: var(--ink);
+      background:
+        radial-gradient(circle at 90% 0%, #d9f6ff 0%, transparent 40%),
+        radial-gradient(circle at 10% 10%, #e3fff4 0%, transparent 30%),
+        var(--bg);
+      line-height: 1.6;
+    }
+    .wrap {
+      max-width: 1050px;
+      margin: 0 auto;
+      padding: 32px 20px 64px;
+    }
+    .hero {
+      background: linear-gradient(125deg, #0f172a 0%, #0b3f63 45%, #0d7f86 100%);
+      color: #f8fafc;
+      border-radius: 20px;
+      padding: 28px 28px 20px;
+      box-shadow: 0 22px 55px rgba(2, 20, 38, 0.35);
+    }
+    .hero h1 { margin: 0 0 10px; font-size: clamp(26px, 4vw, 38px); }
+    .hero p { margin: 0; max-width: 880px; color: #dceafd; }
+    .chips { margin-top: 14px; display: flex; flex-wrap: wrap; gap: 10px; }
+    .chip {
+      background: rgba(255,255,255,0.14);
+      border: 1px solid rgba(255,255,255,0.24);
+      border-radius: 999px;
+      padding: 6px 12px;
+      font-size: 13px;
+    }
+    section {
+      background: var(--panel);
+      border: 1px solid var(--line);
+      border-radius: 16px;
+      margin-top: 16px;
+      padding: 20px;
+    }
+    h2 { margin: 0 0 12px; font-size: 22px; }
+    h3 { margin: 18px 0 8px; font-size: 18px; }
+    p, li { color: var(--muted); }
+    ul { margin: 8px 0 0 20px; padding: 0; }
+    code, pre {
+      font-family: "Cascadia Code", Consolas, Menlo, monospace;
+      font-size: 13px;
+    }
+    pre {
+      margin: 10px 0 0;
+      background: #0f172a;
+      color: #d9ecff;
+      border-radius: 12px;
+      padding: 12px;
+      overflow: auto;
+      border: 1px solid #263142;
+    }
+    .cards {
+      display: grid;
+      grid-template-columns: repeat(auto-fit, minmax(230px, 1fr));
+      gap: 12px;
+    }
+    .card {
+      border: 1px solid var(--line);
+      border-radius: 14px;
+      padding: 14px;
+      background: linear-gradient(180deg, #fbfeff 0%, #f7fbff 100%);
+    }
+    .badge {
+      display: inline-block;
+      padding: 4px 8px;
+      border-radius: 999px;
+      font-size: 12px;
+      color: #065f46;
+      background: #d1fae5;
+      border: 1px solid #a7f3d0;
+    }
+    a { color: #0369a1; text-decoration: none; }
+    a:hover { text-decoration: underline; }
+    .ok { color: var(--ok); font-weight: 600; }
+  </style>
+</head>
+<body>
+  <main class="wrap">
+    <header class="hero">
+      <h1>Gitea Issue DevOps Agent</h1>
+      <p>
+        一个把 <strong>Issue → Branch → Preview Env → 测试闭环</strong> 固化到技能与脚本中的交付方案。
+        核心目标是提升交付速度，同时避免“每个分支都全量起服务”的资源浪费。
+      </p>
+      <div class="chips">
+        <span class="chip">自动 / 半自动 / 全人工</span>
+        <span class="chip">Issue 图片证据抓取</span>
+        <span class="chip">变更范围智能部署</span>
+        <span class="chip">槽位池自动回收</span>
+      </div>
+    </header>
+
+    <section>
+      <h2>核心价值</h2>
+      <div class="cards">
+        <article class="card">
+          <h3>1. 分支隔离提测</h3>
+          <p>每个 issue 绑定分支与预览槽位，主干环境保持稳定，避免相互覆盖。</p>
+        </article>
+        <article class="card">
+          <h3>2. 资源按需分配</h3>
+          <p>根据变更范围判断 <code>client_only/server_only/full_stack</code>，不变更服务端就不重启服务端。</p>
+        </article>
+        <article class="card">
+          <h3>3. 可审计闭环</h3>
+          <p>每次提测都可回溯到 commit、测试结果、环境 URL、验证步骤，且合并始终由工程师人工确认。</p>
+        </article>
+      </div>
+    </section>
+
+    <section>
+      <h2>安装指南</h2>
+      <h3>1) 获取技能仓库</h3>
+      <pre><code>git clone https://fun-md.com/Fun_MD/devops-skills.git
+cd devops-skills</code></pre>
+      <h3>2) 安装到 Codex skills</h3>
+      <pre><code># Linux / macOS
+mkdir -p ~/.codex/skills
+cp -r skills/gitea-issue-devops-agent ~/.codex/skills/
+
+# Windows PowerShell
+New-Item -ItemType Directory -Force $HOME\.codex\skills | Out-Null
+Copy-Item .\skills\gitea-issue-devops-agent $HOME\.codex\skills\gitea-issue-devops-agent -Recurse -Force</code></pre>
+      <h3>3) 首次引导参数</h3>
+      <ul>
+        <li><code>repo_url</code>（仓库地址）</li>
+        <li><code>api_key</code>（具备 issue 读写权限）</li>
+        <li><code>mode</code>：<code>automatic</code> / <code>semi-automatic</code> / <code>manual</code></li>
+        <li>可选：<code>reviewers</code>、<code>test_entry</code>、<code>deploy_env</code>、<code>health_endpoint</code>、<code>min_quality_score</code></li>
+      </ul>
+    </section>
+
+    <section>
+      <h2>工具使用说明</h2>
+      <h3>issue_audit.py（拉取 issue + 图片证据）</h3>
+      <pre><code>python skills/gitea-issue-devops-agent/scripts/issue_audit.py \
+  --base-url https://fun-md.com \
+  --repo FunMD/document-collab \
+  --token &lt;GITEA_TOKEN&gt; \
+  --state all \
+  --download-attachments \
+  --output-dir .tmp/issue-audit</code></pre>
+
+      <h3>change_scope.py（按改动范围决策部署）</h3>
+      <pre><code>python skills/gitea-issue-devops-agent/scripts/change_scope.py \
+  --repo-path . \
+  --base-ref origin/main \
+  --head-ref HEAD</code></pre>
+
+      <h3>preview_slot_allocator.py（分配 / 复用 / 释放槽位）</h3>
+      <pre><code>python skills/gitea-issue-devops-agent/scripts/preview_slot_allocator.py \
+  --state-file .tmp/preview-slots.json \
+  --slots preview-a,preview-b \
+  --repo FunMD/document-collab \
+  --issue 48 \
+  --branch dev \
+  --ttl-hours 24 \
+  --url-template https://{slot}.qa.example.com \
+  --evict-oldest</code></pre>
+    </section>
+
+    <section>
+      <h2>.gitea/workflows 接入</h2>
+      <p>
+        本仓库已包含示例工作流：<code>.gitea/workflows/issue-branch-preview.yml</code> 与
+        <code>.gitea/workflows/preview-slot-reclaim.yml</code>，用于完成以下自动化链路：
+      </p>
+      <ul>
+        <li>push 到 issue 分支后：自动分配槽位 + 变更范围识别 + 选择性部署</li>
+        <li>issue 关闭 / 定时任务：自动释放或回收过期槽位</li>
+      </ul>
+      <p class="ok">建议先在测试仓库验证工作流变量后再推广到生产仓库。</p>
+    </section>
+
+    <section>
+      <span class="badge">Skill Path</span>
+      <p><a href="../skills/gitea-issue-devops-agent/SKILL.md">skills/gitea-issue-devops-agent/SKILL.md</a></p>
+    </section>
+  </main>
+</body>
+</html>
diff --git a/skills/gitea-issue-devops-agent/SKILL.md b/skills/gitea-issue-devops-agent/SKILL.md
new file mode 100644
index 0000000..d7a347e
--- /dev/null
+++ b/skills/gitea-issue-devops-agent/SKILL.md
@@ -0,0 +1,234 @@
+---
+name: gitea-issue-devops-agent
+description: End-to-end Gitea issue delivery workflow with guided onboarding, branch-scoped preview environments, and resource-aware deployment decisions. Use when tasks involve connecting to Gitea, processing text/image issues, fixing code on issue-specified branches, allocating/reusing test environments per branch, running test submission loops, coordinating review approvals, and closing issues only after verified delivery and engineer-confirmed merge.
+---
+
+# Gitea Issue DevOps Agent
+
+## Mandatory Guided Start
+
+Run this interaction before any coding or issue action:
+
+1. Ask for repository address:
+   - preferred: full URL `https://<host>/<owner>/<repo>`
+   - fallback: `base_url` + `owner/repo`
+2. Ask for API key/token with issue read/write permissions.
+3. Ask user to select mode:
+   - `automatic`
+   - `semi-automatic`
+   - `manual` (non-automatic)
+4. Ask optional defaults:
+   - designated reviewers (for semi-automatic mode)
+   - branch test submission entrypoint (CI command/job)
+   - environment policy:
+     - stable main URL (`main` fixed test env)
+     - optional shared QA URL
+     - preview slot pool (for issue branches), e.g. `preview-a,preview-b`
+     - preview URL template, e.g. `https://{slot}.qa.example.com`
+   - deployment environment + health endpoint
+   - minimum issue quality score (default `70`)
+5. Validate connectivity by running:
+   - `python scripts/issue_audit.py --repo <owner/repo> --base-url <gitea_url> --token <token> --state all --download-attachments --output-dir .tmp/issue-audit`
+6. Initialize preview-slot state (if branch previews enabled):
+   - `python scripts/preview_slot_allocator.py --state-file .tmp/preview-slots.json --slots <slot_csv> --list`
+7. Echo back the selected mode and all gate rules, then start work.
+
+If repository or token is missing/invalid, stop and request correction. Never start development without a successful connectivity check.
+
+## Mode Definitions
+
+### 1) Automatic Mode
+
+- Read issue-specified branch and work on that branch.
+- Implement fix, run checks, push branch, allocate/reuse branch preview env, and trigger branch test submission automatically.
+- Monitor test results and issue feedback, then iterate on the same branch until pass.
+- Close issue only after evidence is complete.
+- Merge is still blocked until an engineer explicitly confirms merge approval.
+
+### 2) Semi-Automatic Mode
+
+- Read issue-specified branch and work on that branch.
+- Implement and push fix.
+- Notify designated reviewer with change summary, risk, and test plan.
+- Wait for explicit human review approval.
+- After approval, allocate/reuse branch preview env, trigger branch test submission and continue loop.
+- Close issue only after evidence is complete.
+- Merge is still blocked until an engineer explicitly confirms merge approval.
+
+### 3) Manual Mode (Non-Automatic)
+
+Require explicit human confirmation before each major action:
+
+- selecting issue
+- confirming target branch
+- applying code changes
+- pushing commits
+- triggering tests/deploy
+- closing/reopening issue
+- executing merge
+
+No autonomous transition is allowed in manual mode.
+
+## Branch-First Rules
+
+- Treat issue-declared branch as the source of truth.
+- Accept branch hints from issue fields/body/comments (example: `branch: feat/login-fix`).
+- If branch is missing or ambiguous, ask user/reporter and pause that issue.
+- Do not silently switch branches.
+- Keep one active issue per branch unless user explicitly approves batching.
+
+## Environment Model (Required)
+
+Always avoid `main` and issue branches overwriting each other.
+
+1. `main` fixed env (stable):
+   - one permanent URL for regression/baseline testing
+2. optional shared QA env:
+   - integration testing across multiple completed branches
+3. issue preview slot env (ephemeral pool):
+   - small fixed pool (`N` slots, e.g. 2)
+   - one active branch binds to one slot
+   - issue comments must include slot + URL + branch
+   - close/merge/TTL expiry releases slot
+
+Never deploy different branches to the same fixed URL unless user explicitly approves override.
+
+## Issue -> Branch -> Environment Binding
+
+- Binding key: `<repo>#<issue>#<branch>`
+- Environment selection:
+  - if branch already has assigned slot: reuse same slot
+  - else allocate free slot from pool
+  - if no free slot:
+    - in `automatic`: evict oldest expired/inactive slot if policy allows
+    - in `semi-automatic` / `manual`: request explicit confirmation before eviction
+- Persist slot state in `.tmp/preview-slots.json` via `scripts/preview_slot_allocator.py`
+
+## Resource-Aware Deployment Strategy (Required)
+
+Before every branch test submission, detect change scope:
+
+- `python scripts/change_scope.py --repo-path <local_repo> --base-ref <target_base> --head-ref <branch_or_sha>`
+
+Use the scope result to minimize resource usage:
+
+1. `skip` (docs/tests/chore-only):
+   - do not deploy
+   - post no-op verification evidence
+2. `client_only`:
+   - build/deploy client only
+   - reuse existing shared/stable server
+   - do not start a dedicated server for this branch
+3. `server_only`:
+   - deploy/restart server only
+   - keep existing client if unchanged
+4. `full_stack`:
+   - deploy both client and server
+5. `infra_only`:
+   - apply infra/workflow changes; restart only required components
+
+Hard rule:
+- If server-related scope is unchanged, do not provision/restart dedicated server processes for that issue branch.
+
+## Standard Workflow (All Modes)
+
+### 1) Intake and Prioritization
+
+- Pull issues, comments, and attachments from Gitea API.
+- If issue text/comments indicate image evidence but `attachments_downloaded` is `0`, stop and report image-intake failure before coding.
+- Prioritize in this order:
+  - `closed_but_unresolved`
+  - `open` + `quality_score >= min_quality_score`
+  - `open` + `quality_score < min_quality_score` (request details first)
+  - `closed_open_reopen_candidates`
+- For issues with images, inspect attachments before coding.
+
+### 2) Deduplication and Quality Gate
+
+- Group issues by semantic intent, not literal wording.
+- Keep one parent issue for implementation.
+- Use `references/triage-standard.md` for score and comment templates.
+- For low-quality issues, request details and mark as `needs-info`.
+
+### 3) Fix Execution
+
+- Prefer small, reversible patches.
+- Link every code change to issue ID in commit or PR/MR notes.
+- Split cross-cutting work into incremental commits.
+
+### 4) Verification Gate
+
+- Required:
+  - build/compile passes
+  - affected unit/integration tests pass
+  - smoke path for reported scenario passes
+- For UI/image issues:
+  - compare before/after screenshots
+  - verify in at least one Chromium browser
+
+### 5) Branch Test Submission ("提测")
+
+- Submit testing on the issue branch (CI pipeline + branch preview env).
+- Allocate/reuse branch slot before submission.
+- Apply resource-aware deployment decision from change scope.
+- Post evidence in issue comment:
+  - commit SHA
+  - test run URL and result
+  - environment/slot/URL
+  - deployment scope (`skip`/`client_only`/`server_only`/`full_stack`/`infra_only`)
+  - shared backend reused or dedicated backend started
+  - verification steps
+- If fail/reject, iterate on same branch and re-submit.
+
+### 6) Loop Control
+
+- Continue `fix -> test submission -> feedback -> fix` until done.
+- Reopen immediately if verification fails or regression appears.
+- Do not close based on title-only or assumption-only validation.
+
+### 7) Closure Rule
+
+Close issue only when all are true:
+
+- root cause identified
+- fix verified with reproducible evidence
+- test submission passed
+- closure comment includes commit/test/deploy evidence
+
+### 8) Merge Rule (Always Human-Confirmed)
+
+- Final merge must be approved by an engineer in all modes.
+- Agent can prepare merge notes/checklist, but must wait for explicit merge confirmation.
+- Merge only after confirmation, then post final release evidence.
+
+### 9) Environment Cleanup
+
+- On issue close/merge:
+  - release preview slot
+  - stop branch-only processes (if any)
+  - keep main/shared env untouched
+- On TTL expiry:
+  - reclaim idle slot automatically (automatic mode) or after confirmation (semi/manual)
+
+## Script Usage
+
+- `scripts/issue_audit.py`: collect issues/comments/attachments, detect duplicates, score quality, detect unresolved/closed-open links, extract issue branch hints, and generate reports.
+  - image intake uses three sources: markdown/html links, payload `assets/attachments` fields, and `/issues/*/assets` API endpoints.
+  - if your Gitea blocks the assets endpoints, pass `--skip-asset-endpoints` and rely on payload extraction.
+- `scripts/preview_slot_allocator.py`: allocate/reuse/release/list preview slots by issue+branch.
+  - allocate example:
+    - `python scripts/preview_slot_allocator.py --state-file .tmp/preview-slots.json --slots preview-a,preview-b --repo <owner/repo> --issue 48 --branch dev --ttl-hours 24 --url-template https://{slot}.qa.example.com`
+  - release example:
+    - `python scripts/preview_slot_allocator.py --state-file .tmp/preview-slots.json --slots preview-a,preview-b --release --repo <owner/repo> --issue 48 --branch dev`
+- `scripts/change_scope.py`: detect changed scope and recommend minimum deploy strategy.
+  - `python scripts/change_scope.py --repo-path <repo> --base-ref origin/main --head-ref HEAD`
+- `references/triage-standard.md`: scoring rubric and templates for needs-info, review request, test submission, and merge approval.
+
+## Operational Constraints
+
+- Never bulk-close issues without per-issue verification evidence.
+- Never ignore attachment images for UI/interaction issues.
+- Never merge feature requests and bugfixes into one untraceable commit.
+- Never bypass engineer merge confirmation.
+- Never allow branch previews to overwrite main stable env.
+- Never start dedicated branch server when scope indicates client-only changes.
diff --git a/skills/gitea-issue-devops-agent/agents/openai.yaml b/skills/gitea-issue-devops-agent/agents/openai.yaml
new file mode 100644
index 0000000..6394055
--- /dev/null
+++ b/skills/gitea-issue-devops-agent/agents/openai.yaml
@@ -0,0 +1,4 @@
+interface:
+  display_name: "Gitea Issue DevOps Agent"
+  short_description: "Guided Gitea issue delivery with execution modes, branch preview slots, and resource-aware deployments"
+  default_prompt: "Start with guided setup (repo URL, API key, mode, env policy), process issues on issue-specified branches, bind each branch to preview slots, decide deploy scope from git diff (skip/client-only/server-only/full-stack), avoid dedicated server restarts when backend is unchanged, run fix/test loops, and require engineer confirmation before final merge."
diff --git a/skills/gitea-issue-devops-agent/references/triage-standard.md b/skills/gitea-issue-devops-agent/references/triage-standard.md
new file mode 100644
index 0000000..62b37f1
--- /dev/null
+++ b/skills/gitea-issue-devops-agent/references/triage-standard.md
@@ -0,0 +1,140 @@
+# Triage Standard
+
+## Quality Score (0-100)
+
+- `+20` Expected vs actual is explicit.
+- `+20` Reproduction steps are explicit.
+- `+15` Environment is explicit (browser/device/version).
+- `+15` Attachment exists (image/video/log).
+- `+10` Title is specific enough to infer scope.
+- `+20` Body has enough detail for engineering action.
+
+`pass` = score `>= 70`.
+
+## Status Decision
+
+- `closed_but_unresolved`: issue is closed but has clear reporter feedback indicating “still not fixed”.
+- `closed_open_reopen_candidate`: issue is closed, but a newer open issue has high title/body similarity and should be manually reviewed for mis-close or regression.
+- `ready_for_fix`: open issue, quality pass, reproducible.
+- `needs-info`: open issue, quality below threshold.
+- `duplicate`: semantically same as another issue.
+- `enhancement-epic`: feature scope too large for a single fix cycle; split into sub-issues first.
+
+## Mandatory Review Gates
+
+- `image-first`: when issue body/comments contain screenshots, review the image evidence before coding.
+- `image-retrieval-proof`: include `attachment_urls_detected` and `attachments_downloaded` from audit summary to prove image intake worked.
+- `real-ai-only`: for AI conversation/generation issues, verify against real provider APIs with valid keys. Do not close based on mock-only behavior.
+- `closure-evidence`: close only with commit, test result, deploy proof, and verification path.
+
+## Needs-Info Comment Template
+
+Use this comment when score is below threshold:
+
+```text
+[issue-quality-feedback-v1]
+当前工单暂不满足开发准入标准，已进入待补充队列。请补充以下信息后我们将立即纳入修复流程：
+1) 复现步骤（至少 3 步）
+2) 期望结果 vs 实际结果
+3) 环境信息（浏览器/系统/时间）
+4) 截图或录屏（建议标注异常区域）
+```
+
+## Guided Start Template
+
+Use this at session start before any implementation:
+
+```text
+[devops-startup-check-v1]
+开始处理前请提供：
+1) 仓库地址（完整 URL 或 base_url + owner/repo）
+2) API Key（具备 issue 读写权限）
+3) 执行模式（三选一）：
+   - automatic：自动修复+提测循环，最终合并仍需工程师确认
+   - semi-automatic：修复后先人工 review，再提测循环
+   - manual：全流程人工确认
+可选：指定 reviewer、提测命令、部署环境、健康检查地址。
+可选（推荐）：主环境 URL、共享 QA URL、预览槽位池（如 preview-a/preview-b）和预览 URL 模板。
+```
+
+## Review Request Template (Semi-Automatic)
+
+```text
+[issue-review-request-v1]
+已完成本轮修复，等待指定 reviewer 确认后进入提测：
+- issue: #<number>
+- branch: <branch>
+- commit: <sha>
+- change summary: <summary>
+- risk: <risk notes>
+- test plan: <plan>
+请回复“review-approved”或给出修改意见。
+```
+
+## Test Submission Template
+
+```text
+[issue-test-submit-v1]
+已按分支提测：
+- issue: #<number>
+- branch: <branch>
+- commit: <sha>
+- pipeline/test run: <url>
+- environment: <env/version>
+- preview slot: <slot>
+- preview url: <url>
+- deploy scope: <skip|client_only|server_only|full_stack|infra_only>
+- server strategy: <reused-shared|dedicated-branch-server>
+- verify steps: <steps>
+如失败或结果不符合预期，将继续同分支迭代修复。
+```
+
+## Preview Slot Allocation Template
+
+```text
+[issue-preview-slot-v1]
+已分配提测环境（按 issue+branch 绑定）：
+- issue: #<number>
+- branch: <branch>
+- slot: <preview-a|preview-b|...>
+- preview url: <url>
+- ttl: <hours>
+说明：同一分支会复用该 slot；关闭/合并后自动释放。
+```
+
+## Preview Slot Release Template
+
+```text
+[issue-preview-release-v1]
+已释放提测环境：
+- issue: #<number>
+- branch: <branch>
+- slot: <slot>
+- reason: <merged|closed|ttl-expired|manual-release>
+```
+
+## Merge Approval Template
+
+```text
+[merge-approval-check-v1]
+准备执行最终合并，请工程师确认：
+- issue: #<number>
+- branch: <branch>
+- target: <target branch>
+- review status: <approved/pending>
+- test status: <passed/failed>
+- release evidence: <links>
+请明确回复“merge-approved”后再执行合并。
+```
+
+## Closure Comment Template
+
+```text
+[issue-verified-close-v1]
+已修复并发布。
+- commit: <sha>
+- tests: <summary>
+- deploy: <pipeline/run url>
+- verify: <how verified>
+如仍可复现，请附最新截图和复现步骤，我们将立即 reopen。
+```
diff --git a/skills/gitea-issue-devops-agent/scripts/change_scope.py b/skills/gitea-issue-devops-agent/scripts/change_scope.py
new file mode 100644
index 0000000..ac7135c
--- /dev/null
+++ b/skills/gitea-issue-devops-agent/scripts/change_scope.py
@@ -0,0 +1,184 @@
+#!/usr/bin/env python3
+from __future__ import annotations
+
+import argparse
+import json
+import subprocess
+import sys
+from pathlib import Path
+from typing import Any
+
+DOC_EXTS = {".md", ".rst", ".txt", ".adoc"}
+TEST_MARKERS = ("/tests/", "/test/", "__tests__", ".spec.", ".test.")
+CLIENT_PREFIXES = ("client/", "frontend/", "web/", "ui/")
+SERVER_PREFIXES = ("server/", "backend/", "api/", "services/", "worker/")
+INFRA_PREFIXES = (
+    "deploy/",
+    ".gitea/workflows/",
+    ".github/workflows/",
+    "infra/",
+    "k8s/",
+    "helm/",
+    "nginx/",
+)
+SHARED_RUNTIME_FILES = {
+    "package.json",
+    "package-lock.json",
+    "pnpm-lock.yaml",
+    "yarn.lock",
+    "requirements.txt",
+    "pyproject.toml",
+    "poetry.lock",
+    "Dockerfile",
+    "docker-compose.yml",
+    "docker-compose.yaml",
+}
+
+
+def run_git(cwd: Path, *args: str) -> str:
+    result = subprocess.run(
+        ["git", *args],
+        cwd=str(cwd),
+        check=False,
+        capture_output=True,
+        text=True,
+        encoding="utf-8",
+    )
+    if result.returncode != 0:
+        raise RuntimeError(result.stderr.strip() or f"git {' '.join(args)} failed")
+    return result.stdout.strip()
+
+
+def normalize_path(raw: str) -> str:
+    return raw.replace("\\", "/").lstrip("./")
+
+
+def classify_file(path: str) -> set[str]:
+    categories: set[str] = set()
+    normalized = normalize_path(path)
+    lower = normalized.lower()
+    suffix = Path(lower).suffix
+
+    if lower.startswith("docs/") or suffix in DOC_EXTS:
+        categories.add("docs")
+    if any(marker in lower for marker in TEST_MARKERS):
+        categories.add("tests")
+    if any(lower.startswith(prefix) for prefix in CLIENT_PREFIXES):
+        categories.add("client")
+    if any(lower.startswith(prefix) for prefix in SERVER_PREFIXES):
+        categories.add("server")
+    if any(lower.startswith(prefix) for prefix in INFRA_PREFIXES):
+        categories.add("infra")
+    if Path(lower).name in SHARED_RUNTIME_FILES:
+        categories.add("shared_runtime")
+
+    if not categories:
+        categories.add("unknown")
+    return categories
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(
+        description="Detect branch change scope and recommend minimum deployment strategy."
+    )
+    parser.add_argument("--repo-path", default=".", help="Local git repository path")
+    parser.add_argument("--base-ref", required=True, help="Base ref, e.g. origin/main")
+    parser.add_argument("--head-ref", default="HEAD", help="Head ref, e.g. feature branch or SHA")
+    parser.add_argument("--no-files", action="store_true", help="Do not include changed file list in output")
+    args = parser.parse_args()
+
+    repo_path = Path(args.repo_path).resolve()
+
+    try:
+        merge_base = run_git(repo_path, "merge-base", args.base_ref, args.head_ref)
+        diff_output = run_git(repo_path, "diff", "--name-only", f"{merge_base}..{args.head_ref}")
+    except Exception as error:  # noqa: BLE001
+        print(json.dumps({"error": str(error)}, ensure_ascii=False, indent=2))
+        sys.exit(2)
+
+    changed_files = [normalize_path(line) for line in diff_output.splitlines() if line.strip()]
+    file_classes: list[dict[str, Any]] = []
+    category_union: set[str] = set()
+    for path in changed_files:
+        categories = classify_file(path)
+        category_union.update(categories)
+        file_classes.append({"path": path, "categories": sorted(categories)})
+
+    has_docs = "docs" in category_union
+    has_tests = "tests" in category_union
+    has_client = "client" in category_union
+    has_server = "server" in category_union
+    has_infra = "infra" in category_union
+    has_shared_runtime = "shared_runtime" in category_union
+    has_unknown = "unknown" in category_union
+
+    only_docs_tests = bool(changed_files) and category_union.issubset({"docs", "tests"})
+    scope = "skip"
+    reasons: list[str] = []
+
+    if not changed_files:
+        scope = "skip"
+        reasons.append("no changed files")
+    elif only_docs_tests:
+        scope = "skip"
+        reasons.append("docs/tests-only changes")
+    else:
+        require_server = has_server or has_shared_runtime
+        require_client = has_client or has_shared_runtime
+
+        if has_unknown:
+            scope = "full_stack"
+            reasons.append("unknown file changes detected -> conservative full-stack deploy")
+            require_server = True
+            require_client = True
+        elif has_infra and not require_server and not require_client:
+            scope = "infra_only"
+            reasons.append("infra/workflow-only changes")
+        elif require_client and not require_server:
+            scope = "client_only"
+            reasons.append("client-only changes")
+        elif require_server and not require_client:
+            scope = "server_only"
+            reasons.append("server-only changes")
+        else:
+            scope = "full_stack"
+            reasons.append("client/server/shared runtime changes")
+
+    should_restart_server = scope in {"server_only", "full_stack"}
+    should_restart_client = scope in {"client_only", "full_stack"}
+
+    payload: dict[str, Any] = {
+        "repo_path": str(repo_path.as_posix()),
+        "base_ref": args.base_ref,
+        "head_ref": args.head_ref,
+        "merge_base": merge_base,
+        "changed_files_count": len(changed_files),
+        "scope": scope,
+        "categories": {
+            "docs": has_docs,
+            "tests": has_tests,
+            "client": has_client,
+            "server": has_server,
+            "infra": has_infra,
+            "shared_runtime": has_shared_runtime,
+            "unknown": has_unknown,
+        },
+        "only_docs_tests": only_docs_tests,
+        "actions": {
+            "skip_deploy": scope == "skip",
+            "should_restart_client": should_restart_client,
+            "should_restart_server": should_restart_server,
+            "reuse_shared_server": not should_restart_server,
+            "requires_dedicated_server": should_restart_server,
+        },
+        "reasons": reasons,
+    }
+    if not args.no_files:
+        payload["changed_files"] = changed_files
+        payload["file_classification"] = file_classes
+
+    print(json.dumps(payload, ensure_ascii=False, indent=2))
+
+
+if __name__ == "__main__":
+    main()
diff --git a/skills/gitea-issue-devops-agent/scripts/issue_audit.py b/skills/gitea-issue-devops-agent/scripts/issue_audit.py
new file mode 100644
index 0000000..033f376
--- /dev/null
+++ b/skills/gitea-issue-devops-agent/scripts/issue_audit.py
@@ -0,0 +1,873 @@
+#!/usr/bin/env python3
+from __future__ import annotations
+
+import argparse
+import html
+import hashlib
+import json
+import re
+from dataclasses import dataclass
+from datetime import datetime, timezone
+from difflib import SequenceMatcher
+from pathlib import Path
+from typing import Any
+from urllib.error import HTTPError, URLError
+from urllib.parse import urlencode, urljoin, urlparse
+from urllib.request import Request, urlopen
+
+IMG_MD_RE = re.compile(
+    r"!\[[^\]]*\]\(\s*<?([^\s>)]+)(?:\s+[\"'][^\"']*[\"'])?\s*\)"
+)
+IMG_HTML_RE = re.compile(r"<img[^>]+src=[\"']([^\"']+)[\"']", re.IGNORECASE)
+IMG_URL_RE = re.compile(r"(https?://[^\s)]+?\.(?:png|jpg|jpeg|gif|webp|svg))", re.IGNORECASE)
+ATTACHMENT_PATH_RE = re.compile(
+    r"((?:https?://[^\s)\"'>]+)?/(?:attachments|repo-attachments|api/v1/repos/[^\s)\"'>]+/issues(?:/comments)?/\d+/assets/\d+)[^\s)\"'>]*)",
+    re.IGNORECASE,
+)
+UNRESOLVED_KEYWORDS = (
+    "未修复",
+    "没有修复",
+    "问题还在",
+    "依旧",
+    "仍然",
+    "还是",
+    "无法",
+    "没解决",
+    "still not fixed",
+    "not fixed",
+    "cannot reproduce? no",
+    "failed",
+    "broken",
+)
+QUALITY_MARKER = "[issue-quality-feedback-v1]"
+BRANCH_LABEL_RE = re.compile(
+    r"(?:^|[\r\n])\s*(?:branch|target branch|working branch|fix branch|分支|目标分支)\s*[:：=]\s*`?([A-Za-z0-9._/\-]+)`?",
+    re.IGNORECASE,
+)
+BRANCH_INLINE_RE = re.compile(
+    r"(?:^|[\s,;])(?:/branch|branch)\s+`?([A-Za-z0-9._/\-]+)`?",
+    re.IGNORECASE,
+)
+BRANCH_ALLOWED_RE = re.compile(r"^[A-Za-z0-9._/\-]+$")
+
+
+@dataclass
+class IssueEntry:
+    number: int
+    state: str
+    title: str
+    body: str
+    created_at: str
+    updated_at: str
+    closed_at: str | None
+    comments: list[dict[str, Any]]
+    attachments: list[str]
+    quality_score: int
+    target_branch: str | None
+
+    def brief(self) -> dict[str, Any]:
+        return {
+            "number": self.number,
+            "state": self.state,
+            "title": self.title,
+            "quality_score": self.quality_score,
+            "target_branch": self.target_branch,
+            "attachments": len(self.attachments),
+            "created_at": self.created_at,
+            "updated_at": self.updated_at,
+            "closed_at": self.closed_at,
+        }
+
+
+def _to_datetime(value: str | None) -> datetime | None:
+    raw = (value or "").strip()
+    if not raw:
+        return None
+    try:
+        return datetime.fromisoformat(raw.replace("Z", "+00:00"))
+    except ValueError:
+        return None
+
+
+def _request_json(
+    base_url: str,
+    token: str,
+    path: str,
+    query: dict[str, Any] | None = None,
+    method: str = "GET",
+    body: dict[str, Any] | None = None,
+) -> Any:
+    query_str = f"?{urlencode(query)}" if query else ""
+    url = f"{base_url.rstrip('/')}{path}{query_str}"
+    payload = None if body is None else json.dumps(body).encode("utf-8")
+    req = Request(
+        url,
+        method=method,
+        headers={
+            "Authorization": f"token {token}",
+            "Content-Type": "application/json",
+            "Accept": "application/json",
+        },
+        data=payload,
+    )
+    with urlopen(req, timeout=30) as resp:
+        raw = resp.read().decode("utf-8")
+    return json.loads(raw) if raw else None
+
+
+def _request_binary(url: str, token: str) -> tuple[bytes, str | None]:
+    header_candidates = (
+        {"Authorization": f"token {token}"},
+        {"Authorization": f"Bearer {token}"},
+        {"X-Gitea-Token": token},
+        {"Authorization": f"token {token}", "X-Gitea-Token": token},
+    )
+    last_error: Exception | None = None
+    for auth_headers in header_candidates:
+        req = Request(
+            url,
+            method="GET",
+            headers={
+                "Accept": "*/*",
+                **auth_headers,
+            },
+        )
+        try:
+            with urlopen(req, timeout=30) as resp:
+                content = resp.read()
+                content_type = resp.headers.get("Content-Type")
+            return content, content_type
+        except HTTPError as error:
+            last_error = error
+            if error.code in {401, 403}:
+                continue
+            raise
+        except URLError as error:
+            last_error = error
+            continue
+    if last_error is not None:
+        raise last_error
+    raise RuntimeError("failed to download attachment")
+
+
+def _normalize_url(raw_url: str, base_url: str) -> str | None:
+    candidate = html.unescape(str(raw_url or "").strip())
+    if not candidate:
+        return None
+    candidate = candidate.strip("<>\"'")
+    if not candidate:
+        return None
+    if candidate.startswith("//"):
+        base_scheme = urlparse(base_url).scheme or "https"
+        candidate = f"{base_scheme}:{candidate}"
+    if candidate.startswith("http://") or candidate.startswith("https://"):
+        return candidate
+    return urljoin(f"{base_url.rstrip('/')}/", candidate)
+
+
+def _asset_to_urls(asset: dict[str, Any], base_url: str) -> list[str]:
+    urls: list[str] = []
+    for key in ("browser_download_url", "download_url", "url", "href", "link"):
+        normalized = _normalize_url(str(asset.get(key) or ""), base_url)
+        if normalized and normalized not in urls:
+            urls.append(normalized)
+    uuid_value = str(asset.get("uuid") or "").strip()
+    if uuid_value:
+        fallback = _normalize_url(f"/attachments/{uuid_value}", base_url)
+        if fallback and fallback not in urls:
+            urls.append(fallback)
+    return urls
+
+
+def _extract_asset_urls(payload: dict[str, Any], base_url: str) -> list[str]:
+    results: list[str] = []
+    for key in ("assets", "attachments"):
+        assets = payload.get(key) or []
+        if not isinstance(assets, list):
+            continue
+        for asset in assets:
+            if not isinstance(asset, dict):
+                continue
+            for url in _asset_to_urls(asset, base_url):
+                if url not in results:
+                    results.append(url)
+    return results
+
+
+def _request_json_optional(
+    *,
+    base_url: str,
+    token: str,
+    path: str,
+    query: dict[str, Any] | None = None,
+) -> Any | None:
+    try:
+        return _request_json(base_url, token, path, query=query)
+    except HTTPError as error:
+        if error.code in {401, 403, 404, 405}:
+            return None
+        raise
+    except URLError:
+        return None
+
+
+def _list_asset_urls_from_endpoint(
+    *,
+    base_url: str,
+    token: str,
+    path: str,
+) -> list[str]:
+    urls: list[str] = []
+    page = 1
+    while True:
+        payload = _request_json_optional(
+            base_url=base_url,
+            token=token,
+            path=path,
+            query={"limit": 50, "page": page},
+        )
+        if payload is None:
+            break
+        if not isinstance(payload, list) or not payload:
+            break
+        for asset in payload:
+            if not isinstance(asset, dict):
+                continue
+            for url in _asset_to_urls(asset, base_url):
+                if url not in urls:
+                    urls.append(url)
+        if len(payload) < 50:
+            break
+        page += 1
+    return urls
+
+
+def _list_issue_attachment_urls(
+    *,
+    base_url: str,
+    api_root: str,
+    token: str,
+    issue_number: int,
+) -> list[str]:
+    return _list_asset_urls_from_endpoint(
+        base_url=base_url,
+        token=token,
+        path=f"{api_root}/issues/{issue_number}/assets",
+    )
+
+
+def _list_comment_attachment_urls(
+    *,
+    base_url: str,
+    api_root: str,
+    token: str,
+    comment_id: int,
+) -> list[str]:
+    return _list_asset_urls_from_endpoint(
+        base_url=base_url,
+        token=token,
+        path=f"{api_root}/issues/comments/{comment_id}/assets",
+    )
+
+
+def _extract_attachments(text: str, base_url: str) -> list[str]:
+    if not text:
+        return []
+    urls = [
+        *IMG_MD_RE.findall(text),
+        *IMG_HTML_RE.findall(text),
+        *IMG_URL_RE.findall(text),
+        *ATTACHMENT_PATH_RE.findall(text),
+    ]
+    normalized: list[str] = []
+    for url in urls:
+        cleaned = _normalize_url(str(url), base_url)
+        if cleaned:
+            normalized.append(cleaned)
+    return sorted(set(normalized))
+
+
+def _normalize_branch_name(raw_value: str) -> str | None:
+    candidate = str(raw_value or "").strip().strip("`'\"")
+    candidate = re.sub(r"[),.;]+$", "", candidate)
+    if not candidate:
+        return None
+    if len(candidate) > 160:
+        return None
+    if not BRANCH_ALLOWED_RE.fullmatch(candidate):
+        return None
+    return candidate
+
+
+def _extract_branch_hints(text: str) -> list[str]:
+    if not text:
+        return []
+    results: list[str] = []
+    for regex in (BRANCH_LABEL_RE, BRANCH_INLINE_RE):
+        for match in regex.findall(text):
+            branch = _normalize_branch_name(match)
+            if branch and branch not in results:
+                results.append(branch)
+    return results
+
+
+def _pick_issue_branch(body: str, comments: list[dict[str, Any]]) -> str | None:
+    for branch in _extract_branch_hints(body):
+        return branch
+    for comment in reversed(comments):
+        for branch in _extract_branch_hints(str(comment.get("body") or "")):
+            return branch
+    return None
+
+
+def _normalize_for_similarity(text: str) -> str:
+    lowered = text.lower()
+    lowered = re.sub(r"[`*_>#~=\[\](){}:;,.!?/\\|+-]+", " ", lowered)
+    lowered = re.sub(r"\s+", " ", lowered).strip()
+    return lowered
+
+
+def _quality_score(issue: dict[str, Any], attachments: list[str], comments: list[dict[str, Any]]) -> int:
+    title = str(issue.get("title") or "")
+    body = str(issue.get("body") or "")
+    comment_blob = "\n".join(str(item.get("body") or "") for item in comments[:5])
+    text = f"{title}\n{body}\n{comment_blob}"
+
+    score = 0
+    if re.search(
+        r"(期望|expected).{0,24}(实际|actual)|(实际|actual).{0,24}(期望|expected)",
+        text,
+        re.I | re.S,
+    ):
+        score += 20
+    if re.search(r"(复现|步骤|step|how to reproduce|重现)", text, re.I):
+        score += 20
+    if re.search(r"(浏览器|browser|系统|os|版本|version|设备|device|时间)", text, re.I):
+        score += 15
+    if attachments:
+        score += 15
+    if len(title.strip()) >= 6:
+        score += 10
+    if len(re.sub(r"\s+", "", body)) >= 40:
+        score += 20
+    return min(100, score)
+
+
+def _contains_unresolved_feedback(comments: list[dict[str, Any]]) -> bool:
+    for comment in comments:
+        body = str(comment.get("body") or "").lower()
+        if any(keyword in body for keyword in UNRESOLVED_KEYWORDS):
+            return True
+    return False
+
+
+def _issue_similarity(left: IssueEntry, right: IssueEntry) -> float:
+    lhs = _normalize_for_similarity(f"{left.title} {left.body[:700]}")
+    rhs = _normalize_for_similarity(f"{right.title} {right.body[:700]}")
+    if not lhs or not rhs:
+        return 0.0
+    return SequenceMatcher(None, lhs, rhs).ratio()
+
+
+def _title_ngrams(title: str) -> set[str]:
+    normalized = re.sub(r"\s+", "", title.lower())
+    normalized = re.sub(r"[^a-z0-9\u4e00-\u9fff]", "", normalized)
+    grams: set[str] = set()
+    for size in (2, 3):
+        for idx in range(len(normalized) - size + 1):
+            gram = normalized[idx : idx + size]
+            if not gram or gram.isdigit():
+                continue
+            grams.add(gram)
+    return grams
+
+
+def _build_duplicate_groups(entries: list[IssueEntry], threshold: float) -> list[list[int]]:
+    if not entries:
+        return []
+
+    pairs: list[tuple[int, int]] = []
+    numbers = [item.number for item in entries]
+    for i in range(len(entries)):
+        for j in range(i + 1, len(entries)):
+            ratio = _issue_similarity(entries[i], entries[j])
+            if ratio >= threshold:
+                pairs.append((entries[i].number, entries[j].number))
+
+    groups: list[list[int]] = []
+    seen: set[int] = set()
+    graph: dict[int, set[int]] = {}
+    for a, b in pairs:
+        graph.setdefault(a, set()).add(b)
+        graph.setdefault(b, set()).add(a)
+
+    for number in numbers:
+        if number in seen or number not in graph:
+            continue
+        stack = [number]
+        group: list[int] = []
+        while stack:
+            node = stack.pop()
+            if node in seen:
+                continue
+            seen.add(node)
+            group.append(node)
+            stack.extend(graph.get(node, set()))
+        if len(group) > 1:
+            groups.append(sorted(group))
+    return sorted(groups, key=lambda item: item[0])
+
+
+def _build_closed_open_links(
+    closed_entries: list[IssueEntry],
+    open_entries: list[IssueEntry],
+    threshold: float,
+    min_title_ngram_overlap: int,
+) -> list[dict[str, Any]]:
+    links: list[dict[str, Any]] = []
+    for closed_issue in closed_entries:
+        closed_at = _to_datetime(closed_issue.closed_at) or _to_datetime(closed_issue.updated_at)
+        if not closed_at:
+            continue
+
+        best_open: IssueEntry | None = None
+        best_ratio = 0.0
+        best_overlap = 0
+        closed_grams = _title_ngrams(closed_issue.title)
+        for open_issue in open_entries:
+            open_created = _to_datetime(open_issue.created_at)
+            if open_created and open_created < closed_at:
+                continue
+            ratio = _issue_similarity(closed_issue, open_issue)
+            overlap = len(closed_grams & _title_ngrams(open_issue.title))
+            if ratio > best_ratio or (ratio == best_ratio and overlap > best_overlap):
+                best_ratio = ratio
+                best_overlap = overlap
+                best_open = open_issue
+
+        if (
+            best_open
+            and best_ratio >= threshold
+            and best_overlap >= max(1, min_title_ngram_overlap)
+        ):
+            links.append(
+                {
+                    "closed_issue": closed_issue.number,
+                    "open_issue": best_open.number,
+                    "similarity": round(best_ratio, 4),
+                    "title_ngram_overlap": best_overlap,
+                }
+            )
+    return sorted(links, key=lambda item: item["closed_issue"])
+
+
+def _load_issues(
+    base_url: str,
+    api_root: str,
+    token: str,
+    state: str,
+    *,
+    fetch_asset_endpoints: bool,
+) -> list[IssueEntry]:
+    states = ["open", "closed"] if state == "all" else [state]
+    collected: list[IssueEntry] = []
+    for target_state in states:
+        page = 1
+        while True:
+            issues = _request_json(
+                base_url,
+                token,
+                f"{api_root}/issues",
+                query={"state": target_state, "limit": 50, "page": page},
+            )
+            if not issues:
+                break
+            for issue in issues:
+                number = int(issue["number"])
+                body = str(issue.get("body") or "")
+                comments = _request_json(
+                    base_url,
+                    token,
+                    f"{api_root}/issues/{number}/comments",
+                    query={"limit": 100},
+                )
+                comments = comments or []
+                attachments = _extract_attachments(body, base_url)
+                attachments.extend(_extract_asset_urls(issue, base_url))
+                if fetch_asset_endpoints:
+                    attachments.extend(
+                        _list_issue_attachment_urls(
+                            base_url=base_url,
+                            api_root=api_root,
+                            token=token,
+                            issue_number=number,
+                        )
+                    )
+                for comment in comments:
+                    attachments.extend(_extract_attachments(str(comment.get("body") or ""), base_url))
+                    attachments.extend(_extract_asset_urls(comment, base_url))
+                    comment_id_raw = comment.get("id")
+                    comment_id = (
+                        comment_id_raw
+                        if isinstance(comment_id_raw, int)
+                        else int(comment_id_raw)
+                        if isinstance(comment_id_raw, str) and comment_id_raw.isdigit()
+                        else None
+                    )
+                    if fetch_asset_endpoints and comment_id is not None:
+                        attachments.extend(
+                            _list_comment_attachment_urls(
+                                base_url=base_url,
+                                api_root=api_root,
+                                token=token,
+                                comment_id=comment_id,
+                            )
+                        )
+                attachments = sorted(set(attachments))
+                collected.append(
+                    IssueEntry(
+                        number=number,
+                        state=str(issue.get("state") or target_state),
+                        title=str(issue.get("title") or ""),
+                        body=body,
+                        created_at=str(issue.get("created_at") or ""),
+                        updated_at=str(issue.get("updated_at") or ""),
+                        closed_at=issue.get("closed_at"),
+                        comments=comments,
+                        attachments=attachments,
+                        quality_score=_quality_score(issue, attachments, comments),
+                        target_branch=_pick_issue_branch(body, comments),
+                    )
+                )
+            if len(issues) < 50:
+                break
+            page += 1
+    return sorted(collected, key=lambda item: item.number)
+
+
+def _needs_quality_feedback(issue: IssueEntry, min_score: int) -> bool:
+    if issue.state != "open" or issue.quality_score >= min_score:
+        return False
+    for comment in issue.comments:
+        if QUALITY_MARKER in str(comment.get("body") or ""):
+            return False
+    return True
+
+
+def _quality_feedback_message() -> str:
+    return (
+        f"{QUALITY_MARKER}\n"
+        "当前工单暂不满足开发准入标准，已进入待补充队列。请补充以下信息后我们将立即纳入修复流程：\n"
+        "1) 复现步骤（至少 3 步）\n"
+        "2) 期望结果 vs 实际结果\n"
+        "3) 环境信息（浏览器/系统/时间）\n"
+        "4) 截图或录屏（建议标注异常区域）"
+    )
+
+
+def _pick_ext_from_url_or_mime(url: str, content_type: str | None) -> str:
+    parsed = urlparse(url)
+    suffix = Path(parsed.path).suffix.lower().strip()
+    if suffix in {".png", ".jpg", ".jpeg", ".gif", ".webp", ".svg"}:
+        return suffix
+
+    normalized = (content_type or "").lower()
+    if "png" in normalized:
+        return ".png"
+    if "jpeg" in normalized or "jpg" in normalized:
+        return ".jpg"
+    if "gif" in normalized:
+        return ".gif"
+    if "webp" in normalized:
+        return ".webp"
+    if "svg" in normalized:
+        return ".svg"
+    return ".bin"
+
+
+def _download_attachments(
+    *,
+    entries: list[IssueEntry],
+    token: str,
+    output_dir: Path,
+    max_per_issue: int,
+) -> list[dict[str, Any]]:
+    output_dir.mkdir(parents=True, exist_ok=True)
+    downloaded: list[dict[str, Any]] = []
+    seen_hashes: set[str] = set()
+
+    for issue in entries:
+        if not issue.attachments:
+            continue
+        for idx, url in enumerate(issue.attachments[:max_per_issue], start=1):
+            digest = hashlib.sha1(url.encode("utf-8")).hexdigest()
+            if digest in seen_hashes:
+                continue
+            seen_hashes.add(digest)
+            try:
+                blob, content_type = _request_binary(url, token)
+                ext = _pick_ext_from_url_or_mime(url, content_type)
+                file_name = f"issue-{issue.number}-{idx}-{digest[:8]}{ext}"
+                local_path = output_dir / file_name
+                local_path.write_bytes(blob)
+                downloaded.append(
+                    {
+                        "issue": issue.number,
+                        "url": url,
+                        "path": str(local_path.as_posix()),
+                        "size_bytes": len(blob),
+                        "content_type": content_type or "",
+                        "status": "ok",
+                    }
+                )
+            except Exception as error:  # noqa: BLE001
+                downloaded.append(
+                    {
+                        "issue": issue.number,
+                        "url": url,
+                        "path": "",
+                        "size_bytes": 0,
+                        "content_type": "",
+                        "status": "failed",
+                        "error": str(error),
+                    }
+                )
+    return downloaded
+
+
+def _render_report(
+    output_path: Path,
+    *,
+    unresolved_closed: list[dict[str, Any]],
+    low_quality_open: list[IssueEntry],
+    duplicate_groups: list[list[int]],
+    closed_open_links: list[dict[str, Any]],
+    downloaded_attachments: list[dict[str, Any]],
+    all_entries: list[IssueEntry],
+) -> None:
+    lines: list[str] = []
+    lines.append("# Issue Audit Report")
+    lines.append("")
+    lines.append(f"- total issues: {len(all_entries)}")
+    lines.append(f"- closed_but_unresolved: {len(unresolved_closed)}")
+    lines.append(f"- open_low_quality: {len(low_quality_open)}")
+    lines.append(f"- duplicate_groups: {len(duplicate_groups)}")
+    lines.append(f"- closed_open_reopen_candidates: {len(closed_open_links)}")
+    issues_with_attachments = [item for item in all_entries if item.attachments]
+    lines.append(f"- issues_with_attachments: {len(issues_with_attachments)}")
+    lines.append(
+        f"- attachment_urls_detected: {sum(len(item.attachments) for item in issues_with_attachments)}"
+    )
+    open_entries = [item for item in all_entries if item.state == "open"]
+    open_with_branch = [item for item in open_entries if item.target_branch]
+    lines.append(f"- open_with_branch_hint: {len(open_with_branch)}/{len(open_entries)}")
+    if downloaded_attachments:
+        ok_count = sum(1 for item in downloaded_attachments if item["status"] == "ok")
+        failed_count = sum(1 for item in downloaded_attachments if item["status"] != "ok")
+        lines.append(f"- attachments_downloaded: {ok_count}/{len(downloaded_attachments)}")
+        lines.append(f"- attachments_download_failed: {failed_count}")
+    lines.append("")
+
+    lines.append("## Closed But Unresolved")
+    if not unresolved_closed:
+        lines.append("- none")
+    else:
+        for item in unresolved_closed:
+            lines.append(
+                f"- #{item['number']} {item['title']} (reason={item['reason']}, related_open={item.get('related_open')}, similarity={item.get('similarity')})"
+            )
+    lines.append("")
+
+    lines.append("## Closed/Open Regression Candidates")
+    if not closed_open_links:
+        lines.append("- none")
+    else:
+        for item in closed_open_links:
+            lines.append(
+                f"- closed #{item['closed_issue']} -> open #{item['open_issue']} (similarity={item['similarity']}, title_overlap={item['title_ngram_overlap']})"
+            )
+    lines.append("")
+
+    lines.append("## Open Low Quality")
+    if not low_quality_open:
+        lines.append("- none")
+    else:
+        for issue in low_quality_open:
+            lines.append(
+                f"- #{issue.number} {issue.title} (score={issue.quality_score}, branch={issue.target_branch or 'missing'}, attachments={len(issue.attachments)})"
+            )
+    lines.append("")
+
+    lines.append("## Open Issue Branch Mapping")
+    if not open_entries:
+        lines.append("- none")
+    else:
+        for issue in open_entries:
+            lines.append(f"- #{issue.number} -> {issue.target_branch or 'missing'}")
+    lines.append("")
+
+    lines.append("## Duplicate Groups (Open)")
+    if not duplicate_groups:
+        lines.append("- none")
+    else:
+        for group in duplicate_groups:
+            lines.append(f"- {', '.join(f'#{num}' for num in group)}")
+    lines.append("")
+
+    if downloaded_attachments:
+        lines.append("## Attachment Download Manifest")
+        for item in downloaded_attachments:
+            lines.append(
+                f"- issue #{item['issue']}: {item['status']} -> {item['path'] or item['url']}"
+            )
+        lines.append("")
+
+    output_path.write_text("\n".join(lines), encoding="utf-8")
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description="Audit Gitea issues for delivery workflow.")
+    parser.add_argument("--base-url", required=True, help="Gitea host, e.g. https://fun-md.com")
+    parser.add_argument("--repo", required=True, help="owner/repo")
+    parser.add_argument("--token", required=True, help="Gitea API token")
+    parser.add_argument("--state", default="all", choices=["open", "closed", "all"])
+    parser.add_argument("--output-dir", default=".tmp/issue-audit")
+    parser.add_argument("--min-quality-score", type=int, default=70)
+    parser.add_argument("--dedupe-threshold", type=float, default=0.62)
+    parser.add_argument("--reopen-similarity-threshold", type=float, default=0.27)
+    parser.add_argument(
+        "--reopen-title-overlap",
+        type=int,
+        default=2,
+        help="Minimum 2/3-char title n-gram overlap for closed/open regression candidates.",
+    )
+    parser.add_argument(
+        "--post-quality-feedback",
+        action="store_true",
+        help="Post needs-info comment for low quality open issues.",
+    )
+    parser.add_argument(
+        "--download-attachments",
+        action="store_true",
+        help="Download image attachments to output-dir/attachments for manual visual review.",
+    )
+    parser.add_argument(
+        "--max-attachments-per-issue",
+        type=int,
+        default=8,
+        help="Limit downloaded attachments per issue to avoid huge sync.",
+    )
+    parser.add_argument(
+        "--skip-asset-endpoints",
+        action="store_true",
+        help="Skip /issues/*/assets API calls and only parse URLs from issue/comment payloads.",
+    )
+    args = parser.parse_args()
+
+    owner, repo_name = args.repo.split("/", 1)
+    api_root = f"/api/v1/repos/{owner}/{repo_name}"
+    entries = _load_issues(
+        args.base_url,
+        api_root,
+        args.token,
+        args.state,
+        fetch_asset_endpoints=not args.skip_asset_endpoints,
+    )
+
+    open_entries = [issue for issue in entries if issue.state == "open"]
+    closed_entries = [issue for issue in entries if issue.state == "closed"]
+    issues_with_attachments = [issue for issue in entries if issue.attachments]
+    open_with_branch = [issue for issue in open_entries if issue.target_branch]
+    open_missing_branch = [issue for issue in open_entries if not issue.target_branch]
+    low_quality_open = [issue for issue in open_entries if issue.quality_score < args.min_quality_score]
+    duplicate_groups = _build_duplicate_groups(open_entries, args.dedupe_threshold)
+    closed_open_links = _build_closed_open_links(
+        closed_entries,
+        open_entries,
+        args.reopen_similarity_threshold,
+        args.reopen_title_overlap,
+    )
+
+    unresolved_closed: list[dict[str, Any]] = []
+
+    for issue in closed_entries:
+        if _contains_unresolved_feedback(issue.comments):
+            unresolved_closed.append(
+                {
+                    "number": issue.number,
+                    "title": issue.title,
+                    "reason": "comment_feedback",
+                }
+            )
+
+    unresolved_closed = sorted(unresolved_closed, key=lambda item: item["number"])
+
+    if args.post_quality_feedback:
+        for issue in low_quality_open:
+            if not _needs_quality_feedback(issue, args.min_quality_score):
+                continue
+            _request_json(
+                args.base_url,
+                args.token,
+                f"{api_root}/issues/{issue.number}/comments",
+                method="POST",
+                body={"body": _quality_feedback_message()},
+            )
+
+    output_dir = Path(args.output_dir)
+    output_dir.mkdir(parents=True, exist_ok=True)
+    downloaded_attachments: list[dict[str, Any]] = []
+    if args.download_attachments:
+        downloaded_attachments = _download_attachments(
+            entries=entries,
+            token=args.token,
+            output_dir=output_dir / "attachments",
+            max_per_issue=max(1, args.max_attachments_per_issue),
+        )
+
+    payload = {
+        "summary": {
+            "total": len(entries),
+            "open": len(open_entries),
+            "closed": len(closed_entries),
+            "closed_but_unresolved": len(unresolved_closed),
+            "open_low_quality": len(low_quality_open),
+            "duplicate_groups": len(duplicate_groups),
+            "closed_open_reopen_candidates": len(closed_open_links),
+            "issues_with_attachments": len(issues_with_attachments),
+            "attachment_urls_detected": sum(len(issue.attachments) for issue in issues_with_attachments),
+            "open_with_branch_hint": len(open_with_branch),
+            "open_missing_branch_hint": len(open_missing_branch),
+            "attachments_downloaded": sum(
+                1 for item in downloaded_attachments if item.get("status") == "ok"
+            ),
+            "attachments_download_failed": sum(
+                1 for item in downloaded_attachments if item.get("status") != "ok"
+            ),
+        },
+        "unresolved_closed": unresolved_closed,
+        "closed_open_links": closed_open_links,
+        "open_low_quality": [item.brief() for item in low_quality_open],
+        "open_missing_branch_issues": [item.brief() for item in open_missing_branch],
+        "duplicate_groups": duplicate_groups,
+        "attachments_manifest": downloaded_attachments,
+        "issues": [item.brief() for item in entries],
+    }
+    (output_dir / "issue_audit.json").write_text(
+        json.dumps(payload, ensure_ascii=False, indent=2),
+        encoding="utf-8",
+    )
+    _render_report(
+        output_dir / "issue_audit_report.md",
+        unresolved_closed=unresolved_closed,
+        low_quality_open=low_quality_open,
+        duplicate_groups=duplicate_groups,
+        closed_open_links=closed_open_links,
+        downloaded_attachments=downloaded_attachments,
+        all_entries=entries,
+    )
+    print(json.dumps(payload["summary"], ensure_ascii=False))
+
+
+if __name__ == "__main__":
+    main()
diff --git a/skills/gitea-issue-devops-agent/scripts/preview_slot_allocator.py b/skills/gitea-issue-devops-agent/scripts/preview_slot_allocator.py
new file mode 100644
index 0000000..0ec97b3
--- /dev/null
+++ b/skills/gitea-issue-devops-agent/scripts/preview_slot_allocator.py
@@ -0,0 +1,248 @@
+#!/usr/bin/env python3
+from __future__ import annotations
+
+import argparse
+import json
+import sys
+from datetime import datetime, timedelta, timezone
+from pathlib import Path
+from typing import Any
+
+
+def utc_now() -> datetime:
+    return datetime.now(timezone.utc)
+
+
+def to_iso(value: datetime) -> str:
+    return value.astimezone(timezone.utc).replace(microsecond=0).isoformat()
+
+
+def parse_iso(value: str | None) -> datetime | None:
+    raw = (value or "").strip()
+    if not raw:
+        return None
+    try:
+        return datetime.fromisoformat(raw.replace("Z", "+00:00")).astimezone(timezone.utc)
+    except ValueError:
+        return None
+
+
+def load_state(path: Path) -> dict[str, Any]:
+    if not path.exists():
+        return {"version": 1, "updated_at": to_iso(utc_now()), "allocations": []}
+    try:
+        payload = json.loads(path.read_text(encoding="utf-8"))
+    except json.JSONDecodeError:
+        return {"version": 1, "updated_at": to_iso(utc_now()), "allocations": []}
+    if not isinstance(payload, dict):
+        return {"version": 1, "updated_at": to_iso(utc_now()), "allocations": []}
+    payload.setdefault("version", 1)
+    payload.setdefault("updated_at", to_iso(utc_now()))
+    allocations = payload.get("allocations")
+    if not isinstance(allocations, list):
+        payload["allocations"] = []
+    return payload
+
+
+def save_state(path: Path, payload: dict[str, Any]) -> None:
+    path.parent.mkdir(parents=True, exist_ok=True)
+    payload["updated_at"] = to_iso(utc_now())
+    path.write_text(json.dumps(payload, ensure_ascii=False, indent=2), encoding="utf-8")
+
+
+def parse_slots(raw: str) -> list[str]:
+    slots = [item.strip() for item in raw.split(",") if item.strip()]
+    if not slots:
+        raise ValueError("at least one slot is required")
+    return slots
+
+
+def render_url(slot: str, template: str | None) -> str | None:
+    if not template:
+        return None
+    return template.replace("{slot}", slot)
+
+
+def prune_expired(allocations: list[dict[str, Any]], now: datetime) -> tuple[list[dict[str, Any]], int]:
+    kept: list[dict[str, Any]] = []
+    removed = 0
+    for item in allocations:
+        expires = parse_iso(str(item.get("expires_at") or ""))
+        if expires and expires < now:
+            removed += 1
+            continue
+        kept.append(item)
+    return kept, removed
+
+
+def allocation_sort_key(item: dict[str, Any]) -> datetime:
+    last_seen = parse_iso(str(item.get("last_seen_at") or ""))
+    allocated = parse_iso(str(item.get("allocated_at") or ""))
+    return last_seen or allocated or datetime.fromtimestamp(0, tz=timezone.utc)
+
+
+def output(payload: dict[str, Any]) -> None:
+    print(json.dumps(payload, ensure_ascii=False, indent=2))
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(
+        description="Allocate/reuse/release branch preview slots for issue-driven workflows."
+    )
+    parser.add_argument("--state-file", default=".tmp/preview-slots.json")
+    parser.add_argument("--slots", required=True, help="Comma-separated slot names, e.g. preview-a,preview-b")
+    parser.add_argument("--repo", default="", help="owner/repo")
+    parser.add_argument("--issue", type=int, default=None, help="Issue number")
+    parser.add_argument("--branch", default="", help="Branch name")
+    parser.add_argument("--slot", default="", help="Slot name (optional release filter)")
+    parser.add_argument("--ttl-hours", type=int, default=24, help="Allocation TTL in hours")
+    parser.add_argument("--url-template", default="", help="Optional URL template, use {slot} placeholder")
+    parser.add_argument("--release", action="store_true", help="Release matching allocation(s)")
+    parser.add_argument("--list", action="store_true", help="List active allocations and free slots")
+    parser.add_argument("--evict-oldest", action="store_true", help="Evict oldest allocation when slots are full")
+    args = parser.parse_args()
+
+    state_path = Path(args.state_file)
+    slots = parse_slots(args.slots)
+    state = load_state(state_path)
+
+    now = utc_now()
+    allocations_raw = [item for item in state.get("allocations", []) if isinstance(item, dict)]
+    allocations, pruned_count = prune_expired(allocations_raw, now)
+    state["allocations"] = allocations
+
+    if args.list:
+        used_slots = {str(item.get("slot") or "") for item in allocations}
+        free_slots = [slot for slot in slots if slot not in used_slots]
+        output(
+            {
+                "action": "list",
+                "state_file": str(state_path.as_posix()),
+                "pruned_expired": pruned_count,
+                "total_active": len(allocations),
+                "active_allocations": allocations,
+                "free_slots": free_slots,
+            }
+        )
+        save_state(state_path, state)
+        return
+
+    if args.release:
+        target_repo = args.repo.strip()
+        target_branch = args.branch.strip()
+        target_slot = args.slot.strip()
+        target_issue = args.issue
+
+        kept: list[dict[str, Any]] = []
+        released: list[dict[str, Any]] = []
+        for item in allocations:
+            match = True
+            if target_repo:
+                match = match and str(item.get("repo") or "") == target_repo
+            if target_branch:
+                match = match and str(item.get("branch") or "") == target_branch
+            if target_slot:
+                match = match and str(item.get("slot") or "") == target_slot
+            if target_issue is not None:
+                match = match and int(item.get("issue") or -1) == target_issue
+
+            if match:
+                released.append(item)
+            else:
+                kept.append(item)
+
+        state["allocations"] = kept
+        save_state(state_path, state)
+        output(
+            {
+                "action": "release",
+                "state_file": str(state_path.as_posix()),
+                "released_count": len(released),
+                "released": released,
+                "remaining_count": len(kept),
+            }
+        )
+        return
+
+    repo = args.repo.strip()
+    branch = args.branch.strip()
+    if not repo or not branch:
+        print("allocate requires --repo and --branch", file=sys.stderr)
+        sys.exit(2)
+
+    # Reuse existing allocation for same repo+branch.
+    existing = next(
+        (item for item in allocations if str(item.get("repo") or "") == repo and str(item.get("branch") or "") == branch),
+        None,
+    )
+    if existing is not None:
+        existing["issue"] = args.issue if args.issue is not None else existing.get("issue")
+        existing["last_seen_at"] = to_iso(now)
+        existing["expires_at"] = to_iso(now + timedelta(hours=max(1, args.ttl_hours)))
+        if args.url_template:
+            existing["url"] = render_url(str(existing.get("slot") or ""), args.url_template)
+        save_state(state_path, state)
+        output(
+            {
+                "action": "allocate",
+                "reused": True,
+                "state_file": str(state_path.as_posix()),
+                "allocation": existing,
+                "pruned_expired": pruned_count,
+            }
+        )
+        return
+
+    used_slots = {str(item.get("slot") or "") for item in allocations}
+    free_slots = [slot for slot in slots if slot not in used_slots]
+    evicted: dict[str, Any] | None = None
+
+    if not free_slots:
+        if not args.evict_oldest:
+            output(
+                {
+                    "action": "allocate",
+                    "reused": False,
+                    "allocated": False,
+                    "reason": "no_free_slots",
+                    "state_file": str(state_path.as_posix()),
+                    "active_allocations": allocations,
+                }
+            )
+            sys.exit(3)
+        oldest = sorted(allocations, key=allocation_sort_key)[0]
+        evicted = oldest
+        allocations.remove(oldest)
+        free_slots = [str(oldest.get("slot") or "")]
+
+    slot = free_slots[0]
+    allocation = {
+        "repo": repo,
+        "issue": args.issue,
+        "branch": branch,
+        "slot": slot,
+        "env_id": slot,
+        "url": render_url(slot, args.url_template),
+        "allocated_at": to_iso(now),
+        "last_seen_at": to_iso(now),
+        "expires_at": to_iso(now + timedelta(hours=max(1, args.ttl_hours))),
+        "status": "active",
+    }
+    allocations.append(allocation)
+    state["allocations"] = allocations
+    save_state(state_path, state)
+    output(
+        {
+            "action": "allocate",
+            "reused": False,
+            "allocated": True,
+            "state_file": str(state_path.as_posix()),
+            "allocation": allocation,
+            "evicted": evicted,
+            "pruned_expired": pruned_count,
+        }
+    )
+
+
+if __name__ == "__main__":
+    main()