Release v1.6.0: continuous product cycles and real previews

.github/workflows/auto-company-runtime-ci.yml

@@ -184,6 +184,10 @@ jobs:
         run: python3 scripts/ci/run.py shell-contracts-5 bash tests/test_auto_loop_integration.sh
       - name: Verify real frontend script against deterministic API fixtures
         run: python3 scripts/ci/run.py shell-contracts-6 node tests/test_dashboard_frontend.js
+      - name: Verify published example logic and syntax
+        run: |
+          python3 scripts/ci/run.py showcase-core node --test projects/scopefence/tests/scopefence-core.test.mjs projects/text-meter/tests/counts.test.js
+          python3 scripts/ci/run.py showcase-syntax node --check projects/scope-sheet/app.js
       - name: Save failure evidence
         if: failure()
         uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
@@ -209,11 +213,17 @@ jobs:
           package-manager-cache: false
       - name: Install browser test dependencies
         run: python3 scripts/ci/run.py browser-dependencies npm ci --prefix tests/browser --ignore-scripts --no-audit --no-fund
+      - name: Install product screenshot dependencies
+        run: python3 scripts/ci/run.py media-dependencies npm ci --prefix scripts/media --ignore-scripts --no-audit --no-fund
       - name: Install Chromium
         working-directory: tests/browser
         run: python3 ../../scripts/ci/run.py chromium-install npx playwright install --with-deps chromium
       - name: Test Dashboard with a real browser and isolated server
         run: python3 scripts/ci/run.py dashboard-browser npm test --prefix tests/browser
+      - name: Verify automatic product screenshots and process cleanup
+        env:
+          AUTO_COMPANY_TEST_PRODUCT_MEDIA_BROWSER: '1'
+        run: python3 scripts/ci/run.py product-media python3 -m unittest discover -s tests -p test_product_media.py -v
       - name: Save failure evidence
         if: failure()
         uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1

.gitignore

@@ -140,6 +140,7 @@ logs/
 .auto-loop-budget-paused.*
 .auto-loop.env
 .auto-company.local
+.auto-company/
 .auto-company.local.*
 .auto-company-migrations/
 .auto-loop-awake.pid
@@ -160,6 +161,9 @@ projects/*
 # Explicitly published product examples; autonomous output stays ignored.
 !projects/tabledelta/
 !projects/cuecheck/
+!projects/scopefence/
+!projects/text-meter/
+!projects/scope-sheet/
 
 # Local/private outputs (do not publish)
 AGENTS.md
@@ -212,6 +216,8 @@ docs/*/*
 !docs/windows-setup.md
 !docs/usage-governance.md
 !docs/runtime-observability.md
+!docs/product-media.md
+!docs/product-cycles.md
 # Ignore all memories (keep folder marker only)
 memories/*
 !memories/.gitkeep

PROMPT.md

@@ -66,13 +66,20 @@
 
 ## 收敛规则（强制）
 
+下列 Cycle 1/2/3 是**新探索任务**的收敛顺序，不是每次进程启动后的业务重置。已有产品或已记录的探索任务必须接续共识中的实际进展；程序提供的产品轮次是累计工作记录，不代表业务阶段，也不规定交付必须经历多少轮。停止、重启或切换模型后，不因本次启动计数为 1 而重新选题。
+
 1. **Cycle 1**：Brainstorm，每个 agent 提一个想法，结束时排出 top 3
 2. **Cycle 2**：选 #1，critic-munger 做 Pre-Mortem，research-thompson 验证市场，cfo-campbell 算账。给出 GO / NO-GO
 3. **Cycle 3+**：GO → 建 repo 开始写代码，禁止继续讨论。NO-GO → 试 #2，全不行就强选一个做
 4. **Cycle 2 之后每轮必须产出实物**（文件、repo、部署），纯讨论禁止
 5. **同一个 Next Action 连续出现 2 轮** → 卡住了，换方向或缩范围直接 ship
 6. **凡是前端交付**（页面、界面、组件、dashboard、marketing site）→ 必须先使用 `frontend-design.md`，确保视觉与交互质量，不允许用通用默认风格直接输出
 
+## 产品页面与识别素材
+
+- 前端产品优先复用已有的有效 SVG 图标；没有时提供一个简洁的 `icon.svg`，使用 `viewBox`、少量颜色和基础几何形状。产品页标识与 favicon 引用 Web 根目录的 `auto-company-icon.svg`：这是程序在交付或轮次结束时写入的已验证图标，勿手动创建或修改。不要使用脚本、外部资源、嵌入 HTML 或复杂滤镜，不调用专用生图模型。程序会校验输入图标，缺失或无效时使用默认图标；图标问题不阻断功能交付。
+- 程序会在轮次收尾与交付文档登记后截取实际页面。根目录 `index.html` 的静态 Web 产品无需另写截图命令；其他受支持入口按 `docs/product-media.md` 声明固定预览配置及必要的展示步骤。非界面产品声明不适用，不编造页面。截图失败不代表功能检查失败，也不能用概念图代替实拍。
+
 ## 人工治理与项目边界（强制）
 
 1. `Human Overrides` 是人类专属区，必须逐字保留；任何改动都会触发整轮共识回滚并暂停循环。

README-ZH.md

@@ -23,9 +23,24 @@
 
 ## 看板预览
 
-![Auto Company 看板](presentation/dashboard-showcase.png)
+[![Auto Company Dashboard](presentation/dashboard-showcase.png)](presentation/dashboard-showcase.png)
 
-新版看板用纵向时间轴串联各轮次，展示简洁工作记录、检查结果、交付成果、用量与日志。截图使用 Outline Desk 连续三轮的真实记录，以只读归档模式展示。标题和摘要来自模型按固定格式填写的工作记录；运行事实和已接入检查的结果由程序记录。缺失或过期的信息明确标注，已记录用量不等于完整账单。采集约定与支持范围见[运行记录说明](docs/runtime-observability.md)。
+ScopeFence 的真实四轮产品记录：04 展开，03、02、01 逐项收起且全部可见；立项前探索单独保留。看板展示工作汇报、检查、文档、实拍、用量与日志。标题和摘要仍是模型填写的记录，运行事实与支持的检查结果由程序采集；缺失、失败、过期和部分用量明确标注。支持范围见[运行记录](docs/runtime-observability.md)、[连续轮次](docs/product-cycles.md)和[自动实拍](docs/product-media.md)说明。
+
+<table>
+  <tr>
+    <th width="50%">Text Meter · English</th>
+    <th width="50%">范围确认单 / Scope Sheet · 中文</th>
+  </tr>
+  <tr>
+    <td width="50%" valign="top"><a href="presentation/dashboards/text-meter.png"><img src="presentation/dashboards/text-meter.png" alt="Text Meter · English Dashboard" width="100%" /></a></td>
+    <td width="50%" valign="top"><a href="presentation/dashboards/scope-sheet.png"><img src="presentation/dashboards/scope-sheet.png" alt="范围确认单 / Scope Sheet · 中文 Dashboard" width="100%" /></a></td>
+  </tr>
+  <tr>
+    <td width="50%" valign="top">两次启动的两个产品轮次，编号从 01 延续到 02。</td>
+    <td width="50%" valign="top">四个产品尝试；两次模型容量错误保留为失败，后续成功记录照常呈现。</td>
+  </tr>
+</table>
 
 ## 这是什么？
 
@@ -49,28 +64,23 @@ daemon (launchd / systemd --user, 崩溃自重启)
 
 ## 运行产物示例
 
-以下是 Auto-Company 自主运行生成的独立应用。截图展示的是产出应用，不是 Auto-Company 本身或其控制台。
-
-人类仅下达启动指令和交付范围；从产品立项、方案讨论、设计开发到测试交付，均由 Agent 团队自主讨论、决策并执行，过程中无需人工介入。
+以下三个独立产品来自实际运行，中英文 README 展示同一套项目。ScopeFence 与范围确认单由默认流程自主选题；Text Meter 来自一个普通的文本统计需求。人类设置了运行权限、语言与停止边界，发布前进行了定向审查和必要修正，具体来源记录在各项目中。产品截图来自公开源码的实际默认界面；上方 Dashboard 保留原始运行事实，未修改失败或工作汇报。
 
 <table>
   <tr>
-    <th width="50%">行间 / TableDelta</th>
-    <th width="50%">幕检 / CueCheck</th>
+    <th width="33%">ScopeFence · English</th>
+    <th width="33%">Text Meter · English</th>
+    <th width="33%">范围确认单 / Scope Sheet · 中文</th>
   </tr>
   <tr>
-    <td width="50%" valign="top"><a href="projects/tabledelta/docs/images/desktop-result.png"><img src="projects/tabledelta/docs/images/desktop-result.png" alt="行间 / TableDelta：CSV 差异核对结果" width="100%" /></a></td>
-    <td width="50%" valign="top"><a href="projects/cuecheck/docs/images/desktop-result.png"><img src="projects/cuecheck/docs/images/desktop-result.png" alt="幕检 / CueCheck：字幕检查与编辑工作台" width="100%" /></a></td>
+    <td width="33%" valign="top"><a href="presentation/products/scopefence-full.png"><img src="presentation/products/scopefence.png" alt="ScopeFence" width="100%" /></a></td>
+    <td width="33%" valign="top"><a href="presentation/products/text-meter-full.png"><img src="presentation/products/text-meter.png" alt="Text Meter" width="100%" /></a></td>
+    <td width="33%" valign="top"><a href="presentation/products/scope-sheet-full.png"><img src="presentation/products/scope-sheet.png" alt="范围确认单 / Scope Sheet" width="100%" /></a></td>
   </tr>
   <tr>
-    <td width="50%" valign="top">
-      <p>对比两份 CSV，查看新增、删除和修改，并导出变化报告。</p>
-      <p><a href="projects/tabledelta/">查看源码 →</a></p>
-    </td>
-    <td width="50%" valign="top">
-      <p>检查 SRT 字幕的时码、重叠与阅读速度，支持逐条编辑、重检和导出。</p>
-      <p><a href="projects/cuecheck/">查看源码 →</a></p>
-    </td>
+    <td width="33%" valign="top"><p>将范围变更整理成无需登录的确认链接；返回链接是可编辑的沟通副本，不是经过验证的批准记录。</p><p><a href="projects/scopefence/">查看源码 →</a></p></td>
+    <td width="33%" valign="top"><p>在浏览器本地即时统计字符、非空白字符、以空白分隔的词与行数。</p><p><a href="projects/text-meter/">查看源码 →</a></p></td>
+    <td width="33%" valign="top"><p>填写合作范围、修改、交付与报价，检查必填项后复制或下载中文确认单。</p><p><a href="projects/scope-sheet/">查看源码 →</a></p></td>
   </tr>
 </table>
 

README.md

@@ -23,9 +23,24 @@ Optional Cursor and OpenAI-compatible adapters require explicit configuration. S
 
 ## Dashboard Preview
 
-![Auto Company Dashboard](presentation/dashboard-showcase.png)
+[![Auto Company Dashboard](presentation/dashboard-showcase.png)](presentation/dashboard-showcase.png)
 
-The dashboard connects cycles in a vertical timeline, with concise work records, check results, deliverables, usage, and logs. This screenshot shows three real Outline Desk cycles in read-only archive mode. Titles and summaries are structured model reports; runtime facts and supported check results are recorded by the program. Missing or expired evidence stays explicit, and recorded usage is not a complete bill. See the [collection contract and supported tools](docs/runtime-observability.md).
+Four real ScopeFence product cycles: 04 is expanded; 03, 02 and 01 remain individually collapsed and visible. Pre-product exploration is kept separately. The journal connects reports, checks, documents, real previews, usage and logs. Titles and summaries remain model-authored reports; runtime facts and supported check results are collected by the program. Missing, failed, stale and partial evidence stays explicit. See the [recording contract](docs/runtime-observability.md), [continuous cycles](docs/product-cycles.md) and [automatic previews](docs/product-media.md).
+
+<table>
+  <tr>
+    <th width="50%">Text Meter · English</th>
+    <th width="50%">范围确认单 / Scope Sheet · 中文</th>
+  </tr>
+  <tr>
+    <td width="50%" valign="top"><a href="presentation/dashboards/text-meter.png"><img src="presentation/dashboards/text-meter.png" alt="Text Meter · English Dashboard" width="100%" /></a></td>
+    <td width="50%" valign="top"><a href="presentation/dashboards/scope-sheet.png"><img src="presentation/dashboards/scope-sheet.png" alt="范围确认单 / Scope Sheet · 中文 Dashboard" width="100%" /></a></td>
+  </tr>
+  <tr>
+    <td width="50%" valign="top">Two product cycles across two separate starts; numbering continues from 01 to 02.</td>
+    <td width="50%" valign="top">Four product attempts. Two provider-capacity failures remain visible alongside the successful later work.</td>
+  </tr>
+</table>
 
 ## What Is This?
 
@@ -49,28 +64,23 @@ Each cycle is an independent CLI call. `memories/consensus.md` is the only cross
 
 ## Generated Applications
 
-These independent applications were created through autonomous Auto-Company runs. The screenshots show the resulting products, not Auto-Company itself or its dashboard.
-
-Humans supplied only the instruction to start and the delivery scope. From product selection and planning through design, development, testing, and delivery, the Agent team discussed, decided, and executed autonomously, with no human intervention in that process.
+These three local products come from actual runs and appear in both README languages. ScopeFence and Scope Sheet chose their directions through the default workflow; Text Meter came from an ordinary text-counting request. Humans set permissions, language and external run boundaries, then reviewed and made necessary publication fixes documented in each project. Product images show the actual default interfaces of the published source; the Dashboards above retain the original reports and failures.
 
 <table>
   <tr>
-    <th width="50%">行间 / TableDelta</th>
-    <th width="50%">幕检 / CueCheck</th>
+    <th width="33%">ScopeFence · English</th>
+    <th width="33%">Text Meter · English</th>
+    <th width="33%">范围确认单 / Scope Sheet · 中文</th>
   </tr>
   <tr>
-    <td width="50%" valign="top"><a href="projects/tabledelta/docs/images/desktop-result.png"><img src="projects/tabledelta/docs/images/desktop-result.png" alt="TableDelta: CSV comparison results" width="100%" /></a></td>
-    <td width="50%" valign="top"><a href="projects/cuecheck/docs/images/desktop-result.png"><img src="projects/cuecheck/docs/images/desktop-result.png" alt="CueCheck: subtitle checking and editing workspace" width="100%" /></a></td>
+    <td width="33%" valign="top"><a href="presentation/products/scopefence-full.png"><img src="presentation/products/scopefence.png" alt="ScopeFence" width="100%" /></a></td>
+    <td width="33%" valign="top"><a href="presentation/products/text-meter-full.png"><img src="presentation/products/text-meter.png" alt="Text Meter" width="100%" /></a></td>
+    <td width="33%" valign="top"><a href="presentation/products/scope-sheet-full.png"><img src="presentation/products/scope-sheet.png" alt="范围确认单 / Scope Sheet" width="100%" /></a></td>
   </tr>
   <tr>
-    <td width="50%" valign="top">
-      <p>Compare two CSV files, inspect added, removed, and changed records, and export a change report.</p>
-      <p><a href="projects/tabledelta/">View source →</a></p>
-    </td>
-    <td width="50%" valign="top">
-      <p>Check SRT subtitle timing, overlaps, and reading speed; edit individual subtitles, recheck, and export.</p>
-      <p><a href="projects/cuecheck/">View source →</a></p>
-    </td>
+    <td width="33%" valign="top"><p>Turn a scope change into a no-login decision link. The returned link is an editable communication copy, not verified approval.</p><p><a href="projects/scopefence/">View source →</a></p></td>
+    <td width="33%" valign="top"><p>Count characters, non-whitespace characters, whitespace-separated words and lines locally as you type.</p><p><a href="projects/text-meter/">View source →</a></p></td>
+    <td width="33%" valign="top"><p>Prepare a Chinese scope note, check required fields, and copy or download the text for a client conversation.</p><p><a href="projects/scope-sheet/">View source →</a></p></td>
   </tr>
 </table>