sparklkt · sparklkt · Jun 16, 2026
diff --git a/.env.example b/.env.example
@@ -11,6 +11,32 @@ MEDIA_OPS_REVIEW_PORT=8787
 # Optional direct token override. Prefer local review-server generated files.
 MEDIA_OPS_REVIEW_TOKEN=
 
+# X / Twitter API for approved publishing and metric collection.
+# Keep real values only in your local .env or system keychain.
+X_API_KEY=
+X_API_SECRET=
+X_ACCESS_TOKEN=
+X_ACCESS_TOKEN_SECRET=
+X_BEARER_TOKEN=
+
+# Image generation provider used by media-ops asset factories.
+OPENAI_API_KEY=
+OPENAI_IMAGE_MODEL=gpt-image-1
+
+# Video generation provider. Use either Seedance token or Volcengine AK/SK.
+SEEDANCE_API_TOKEN=
+VOLCENGINE_ACCESS_KEY_ID=
+VOLCENGINE_SECRET_ACCESS_KEY=
+
+# Optional Reddit API app for trend ingestion and future approved publishing.
+REDDIT_CLIENT_ID=
+REDDIT_CLIENT_SECRET=
+REDDIT_REFRESH_TOKEN=
+REDDIT_USER_AGENT=botcom-workbench-local/0.1
+
+# Browser profile used only for local, approved domestic-platform workflows.
+MEDIA_OPS_BROWSER_PROFILE_DIR=$HOME/BotCom/browser-profile
+
 # Feishu / Lark custom bot
 FEISHU_BOT_WEBHOOK=
 FEISHU_BOT_SECRET=

diff --git a/README.md b/README.md
@@ -49,6 +49,7 @@ Core operating modules:
   - operating-state snapshot;
   - eight-module one-person company map;
   - local operating adapters for CRM, delivery, revenue, assets, and automation;
+  - Growth Autopilot benchmark panel for social-media operating loops;
   - approval queues and mobile approval links;
   - service health, blockers, next actions, and review suggestions.
 - AI beginner guide explaining agents, skills, models, API keys, context, and approval queues.
@@ -172,6 +173,8 @@ BotCom Workbench only proxies to `127.0.0.1`; it does not accept arbitrary backe
 
 See `docs/INTEGRATION.md`.
 
+For the Growth Autopilot loop and social-media benchmark model, see `docs/GROWTH_AUTOPILOT.md`.
+
 For the broader product model, see `docs/ONE_PERSON_COMPANY_OS.md`.
 
 ## Notifications
@@ -215,6 +218,7 @@ See `docs/SECURITY.md`.
 
 - `docs/ARCHITECTURE.md`
 - `docs/CONFIGURATION.md`
+- `docs/GROWTH_AUTOPILOT.md`
 - `docs/INTEGRATION.md`
 - `docs/ONE_PERSON_COMPANY_OS.md`
 - `docs/PUBLISHING.md`

diff --git a/docs/CONFIGURATION.md b/docs/CONFIGURATION.md
@@ -63,6 +63,37 @@ WECHAT_BOT_WEBHOOK=
 WECHAT_WEBHOOK=
 ```
 
+## Growth Autopilot connectors
+
+These variables are optional until you enable the related connector. Keep real values in local `.env` or the OS keychain; do not commit them.
+
+```text
+# X API
+X_API_KEY=
+X_API_SECRET=
+X_ACCESS_TOKEN=
+X_ACCESS_TOKEN_SECRET=
+X_BEARER_TOKEN=
+
+# Image generation
+OPENAI_API_KEY=
+OPENAI_IMAGE_MODEL=gpt-image-1
+
+# Video generation
+SEEDANCE_API_TOKEN=
+VOLCENGINE_ACCESS_KEY_ID=
+VOLCENGINE_SECRET_ACCESS_KEY=
+
+# Reddit trend ingestion / approved posting
+REDDIT_CLIENT_ID=
+REDDIT_CLIENT_SECRET=
+REDDIT_REFRESH_TOKEN=
+REDDIT_USER_AGENT=botcom-workbench-local/0.1
+
+# Local browser-assisted domestic-platform workflows
+MEDIA_OPS_BROWSER_PROFILE_DIR=$HOME/BotCom/browser-profile
+```
+
 ## Local env files
 
 The notification bridge reads, in order:

diff --git a/docs/GROWTH_AUTOPILOT.md b/docs/GROWTH_AUTOPILOT.md
@@ -0,0 +1,115 @@
+# Growth Autopilot
+
+Growth Autopilot is BotCom Workbench's operating loop for AI-assisted social growth. It is designed for one-person companies that want AI workers to run the heavy work while the owner only reviews high-risk decisions.
+
+The goal is not to clone a scheduler. The loop covers the full operating chain:
+
+```text
+trend signals
+  -> positioning and experiment hypothesis
+  -> platform-specific drafts
+  -> image/video asset factory
+  -> quality and compliance gates
+  -> owner approval queue
+  -> publish-prep or approved publishing connector
+  -> metric capture
+  -> next experiment
+```
+
+## Public benchmark posture
+
+BotCom compares the local workflow against public reference products and workflow patterns:
+
+- Postiz-style open-source AI social scheduling;
+- n8n-style automation graphs and social templates;
+- Buffer-style creator scheduling and analytics;
+- Publer-style bulk scheduling and recurring content workflows.
+
+Those systems are useful references, but most of them are scheduler-centric. BotCom's differentiator is the company-level loop: strategy, agent roles, content matrix, asset generation, quality gate, approval UX, connectors, metrics, and the next experiment are evaluated together.
+
+## Local reports
+
+When a compatible `media-ops` workspace is present, the BotCom OS dashboard reads:
+
+```text
+$BOTCOM_MEDIA_OPS_ROOT/reports/growth_autopilot.json
+$BOTCOM_MEDIA_OPS_ROOT/reports/growth_autopilot.md
+$BOTCOM_MEDIA_OPS_ROOT/research/latest_public_benchmark.md
+```
+
+The desktop UI shows:
+
+- BotCom loop score;
+- best public reference score;
+- margin;
+- capability matrix;
+- missing credentials or login steps;
+- recent loop findings;
+- buttons to rerun the loop or open reports.
+
+The score is an architectural and readiness benchmark. Real growth performance still depends on live credentials, platform permissions, publishing cadence, and 24-hour / 72-hour metric feedback.
+
+## Owner approval model
+
+Growth Autopilot should keep irreversible or reputation-sensitive actions behind owner approval:
+
+- account connection;
+- platform publishing;
+- paid API usage above local thresholds;
+- external messages to customers or communities;
+- monetization offers;
+- policy-sensitive content.
+
+Safe local steps can run automatically:
+
+- research summarization;
+- draft generation;
+- image/video draft generation;
+- formatting;
+- queue creation;
+- quality checks;
+- report generation;
+- metric analysis after import.
+
+## Credential slots
+
+Keep real credentials out of Git. Use local `.env`, the operating system keychain, or an approved secret manager.
+
+Common variables:
+
+```text
+X_API_KEY=
+X_API_SECRET=
+X_ACCESS_TOKEN=
+X_ACCESS_TOKEN_SECRET=
+OPENAI_API_KEY=
+OPENAI_IMAGE_MODEL=gpt-image-1
+SEEDANCE_API_TOKEN=
+VOLCENGINE_ACCESS_KEY_ID=
+VOLCENGINE_SECRET_ACCESS_KEY=
+REDDIT_CLIENT_ID=
+REDDIT_CLIENT_SECRET=
+REDDIT_REFRESH_TOKEN=
+FEISHU_BOT_WEBHOOK=
+WECHAT_WORK_BOT_WEBHOOK=
+MEDIA_OPS_BROWSER_PROFILE_DIR=$HOME/BotCom/browser-profile
+```
+
+## Platform connector stance
+
+- **X**: prefer official API for posting and public metrics.
+- **Reddit**: prefer official OAuth API for trend ingestion and approved posting.
+- **TikTok / Douyin**: prefer official content posting APIs where account permissions allow it; use local browser-assisted steps only when explicitly approved by the operator and allowed by the platform/account context.
+- **Xiaohongshu**: treat direct publishing as a connector slot. Use official partner/open-platform paths when available; otherwise keep publishing in owner-reviewed browser or service-provider workflows.
+
+## Extending the loop
+
+A new connector or generator should expose:
+
+1. `status`: ready, blocked, missing credentials, last successful run;
+2. `inputs`: accepted local files or queue fields;
+3. `actions`: dry-run, prepare, publish, collect metrics;
+4. `evidence`: local logs, request IDs, generated files, metric snapshots;
+5. `risk`: whether owner approval is required.
+
+This keeps the product usable for non-technical operators while preserving auditability for advanced users.
diff --git a/docs/ROADMAP.md b/docs/ROADMAP.md
@@ -15,6 +15,7 @@
 - Installer-friendly first-run setup.
 - In-app environment configuration screen.
 - Optional sample media-ops workspace generator.
+- Growth Autopilot panel: public workflow benchmark, capability matrix, missing-credential checklist, and loop rerun command.
 - Better empty-state guidance when no operating adapters are configured.
 - Adapter contracts for CRM, delivery, revenue, assets, and metrics.
 - Signed and notarized macOS release.

diff --git a/public/app.js b/public/app.js
@@ -1880,8 +1880,9 @@ function hlTerm(text, term) {
 
 // ---------- BotCom OS：一人公司经营控制台 ----------
 const botcomDashboard = {
-  ov: null, data: null, review: null, reviewError: '',
-  async show() {
+  ov: null, data: null, review: null, reviewError: '', focus: '',
+  async show(focus = '') {
+    this.focus = focus;
     if (this.ov) this.ov.remove();
     const ov = document.createElement('div');
     this.ov = ov;
@@ -2177,6 +2178,48 @@ const botcomDashboard = {
       </div>
     </section>`;
   },
+  growthAutopilotSection(autopilot) {
+    if (!autopilot || !autopilot.score) {
+      return `<section class="botcom-card botcom-autopilot-card" id="growth-autopilot-workflow">
+        <div class="botcom-card-title">Growth Autopilot · 公开最佳工作流基准</div>
+        <div class="botcom-sub">尚未生成基准报告。运行 loop 后会对比 Postiz、n8n、Buffer、Publer 等公开方案，并列出 API / Cookie / 指标缺口。</div>
+        <div class="botcom-actions inline">
+          <button class="primary" data-cmd="growth-autopilot">运行 Growth Autopilot Loop</button>
+        </div>
+      </section>`;
+    }
+    const score = autopilot.score || {};
+    const caps = autopilot.capabilities || [];
+    const needs = autopilot.needs_from_owner || [];
+    const loops = autopilot.loops || [];
+    const verdict = score.verdict === 'architecturally_ahead' ? '架构和执行闭环已领先公开 scheduler' : '领先但仍需凭据和真实指标';
+    return `<section class="botcom-card botcom-autopilot-card" id="growth-autopilot-workflow">
+      <div class="botcom-card-title">Growth Autopilot · 公开最佳工作流基准</div>
+      <div class="botcom-score-row">
+        <div><span>BotCom Loop</span><b>${escapeHtml(score.botcom_after_loops ?? '—')}</b></div>
+        <div><span>公开最佳参考</span><b>${escapeHtml(score.best_public_reference ?? '—')}</b></div>
+        <div><span>领先差值</span><b class="${Number(score.margin || 0) > 0 ? 'good' : 'warn'}">${escapeHtml(score.margin ?? '—')}</b></div>
+      </div>
+      <div class="botcom-sub">${escapeHtml(verdict)}。这里比较的是“研究→生成→质检→审核→发布准备→复盘→下一轮实验”的完整经营闭环，不是单个发帖按钮。</div>
+      <div class="botcom-cap-grid">
+        ${caps.slice(0, 12).map((c) => `<div class="botcom-cap">
+          <span>${escapeHtml(c.label || c.id)}</span>
+          <b>${escapeHtml(c.score ?? '')}</b>
+          <em>${escapeHtml(c.next_step || '')}</em>
+        </div>`).join('')}
+      </div>
+      ${needs.length ? `<div class="botcom-card-title sub-title">需要你后续提供 / 登录</div><div class="botcom-needs">
+        ${needs.map((n) => `<div class="botcom-need"><b>${escapeHtml(n.label || n.id)}</b><span>${escapeHtml(n.why || '')}</span><em>${escapeHtml(n.fields || '')}</em></div>`).join('')}
+      </div>` : '<div class="botcom-ok">凭据和连接器已齐，可以进入真实发布 loop。</div>'}
+      <div class="botcom-card-title sub-title">最近 Loop</div>
+      <div class="botcom-list">${loops.slice(-4).map((l) => `<div>Loop ${escapeHtml(l.loop)} · score ${escapeHtml(l.score)} · ${escapeHtml(l.focus)}：${escapeHtml(l.finding)}</div>`).join('')}</div>
+      <div class="botcom-actions inline">
+        <button class="primary" data-cmd="growth-autopilot">重跑 Growth Loop</button>
+        <button data-open-autopilot="md">打开报告</button>
+        <button data-open-autopilot="benchmark">打开竞品基准</button>
+      </div>
+    </section>`;
+  },
   render() {
     const d = this.data || {};
     const body = this.ov && this.ov.querySelector('.botcom-body');
@@ -2185,6 +2228,7 @@ const botcomDashboard = {
     const w = d.workbench || {};
     const lr = m.launchReadiness || {};
     const growth = m.growth || {};
+    const autopilot = m.growthAutopilot || null;
     const queue = m.queue || {};
     const services = m.services || {};
     const op = d.operatingAdapters || {};
@@ -2241,6 +2285,8 @@ const botcomDashboard = {
 
       ${this.moduleMapSection(modules)}
 
+      ${this.growthAutopilotSection(autopilot)}
+
       ${this.queueSection(queue)}
 
       <div class="botcom-two">
@@ -2284,13 +2330,26 @@ const botcomDashboard = {
     });
     body.querySelector('[data-open-workbench]')?.addEventListener('click', () => (w.dashboard ? openWith(w.dashboard, 'default') : toast('AI-Workbench 未配置', true)));
     body.querySelectorAll('[data-cmd]').forEach((b) => b.addEventListener('click', () => this.run(cmds.find((c) => c.id === b.dataset.cmd))));
+    body.querySelectorAll('[data-open-autopilot]').forEach((b) => b.addEventListener('click', () => {
+      const kind = b.dataset.openAutopilot;
+      const p = kind === 'benchmark'
+        ? this.mediaPath('research', 'latest_public_benchmark.md')
+        : this.mediaPath('reports', 'growth_autopilot.md');
+      openWith(p, 'default');
+    }));
     body.querySelectorAll('[data-botcom-day]').forEach((b) => b.addEventListener('click', () => {
       body.querySelectorAll('[data-botcom-day]').forEach((x) => x.classList.toggle('active', x === b));
       body.querySelectorAll('[data-q-day]').forEach((row) => row.classList.toggle('hidden', row.dataset.qDay !== b.dataset.botcomDay));
     }));
     body.querySelectorAll('[data-q-open]').forEach((b) => b.addEventListener('click', () => this.openQueuePath(b.dataset.qOpen, b.dataset.day, b.dataset.platform)));
     body.querySelectorAll('[data-q-copy]').forEach((b) => b.addEventListener('click', () => this.copyQueueText(b.dataset.day, b.dataset.platform)));
     body.querySelectorAll('[data-review-action]').forEach((b) => b.addEventListener('click', () => this.reviewAction(b.dataset.reviewAction, b.dataset.day, b.dataset.platform)));
+    if (this.focus === 'growth') {
+      const target = body.querySelector('#growth-autopilot-workflow');
+      target?.scrollIntoView({ block: 'start', behavior: 'smooth' });
+      target?.classList.add('botcom-focus-pulse');
+      window.setTimeout(() => target?.classList.remove('botcom-focus-pulse'), 1800);
+    }
   },
 };
 
@@ -2707,6 +2766,7 @@ function bindEvents() {
   $('#setup-entry').onclick = () => setupWizard.show();
   $('#skills-entry').onclick = () => skillsView.show();
   $('#botcom-entry').onclick = () => botcomDashboard.show();
+  $('#growth-entry').onclick = () => botcomDashboard.show('growth');
   $('#term-newtab').onclick = () => term.newTab();
   $('#term-max').onclick = () => term.toggleMax();
   $('#term-dock').onclick = () => term.setDock(term.dock === 'bottom' ? 'right' : 'bottom');

diff --git a/public/index.html b/public/index.html
@@ -61,6 +61,10 @@
         <svg width="12" height="12" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><path d="M3 13h8V3H3z"/><path d="M13 21h8V11h-8z"/><path d="M13 3h8v6h-8z"/><path d="M3 21h8v-6H3z"/></svg>
         BotCom OS <span class="entry-badge">老板只审核</span>
       </div>
+      <div id="growth-entry" class="nav-title side-link botcom-growth-entry" title="Growth Autopilot：媒体自动化工作流，公开基准对比、内容生产、质检、审核、发布准备和复盘 loop">
+        <svg width="12" height="12" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><path d="M12 2v6"/><path d="M12 16v6"/><path d="M2 12h6"/><path d="M16 12h6"/><path d="m4.93 4.93 4.24 4.24"/><path d="m14.83 14.83 4.24 4.24"/><path d="m4.93 19.07 4.24-4.24"/><path d="m14.83 9.17 4.24-4.24"/></svg>
+        Growth Autopilot <span class="entry-badge">新工作流</span>
+      </div>
       <div id="usage-sec">
         <div class="nav-title" id="usage-toggle" title="Claude Code / Codex 用量——官方限额窗口 + 本机会话日志，不用开着它们">
           <svg width="12" height="12" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><polyline points="13 2 3 14 12 14 11 22 21 10 12 10 13 2"/></svg>