social-autoposter 1.5.0 → 1.6.0

package/bin/cli.js

@@ -23,6 +23,7 @@ const COPY_TARGETS = [
   'skill',
   'setup',
   'browser-agent-configs',
+  'mcp-servers',
 ];
 
 const ENV_TEMPLATE = `# social-autoposter environment variables
@@ -42,16 +43,20 @@ DATABASE_URL=
 const USER_FILES = new Set(['config.json', '.env', 'SKILL.md']);
 
 // Browser agent config templates -> install path under ~/.claude/browser-agent-configs/
+// twitter-harness replaces the retired twitter-agent (2026-05-19). The harness
+// runs a CDP-driven real Chrome on port 9555 backed by an MCP stdio server at
+// ~/.claude/mcp-servers/browser-harness/server.py. installBrowserHarness()
+// below provisions the supporting bits (uv, browser-harness CLI, mcp pkg).
 const BROWSER_AGENT_CONFIGS = [
-  'twitter-agent-mcp.json',
-  'twitter-agent.json',
   'reddit-agent-mcp.json',
   'reddit-agent.json',
   'linkedin-agent-mcp.json',
   'linkedin-agent.json',
+  'twitter-harness-mcp.json',
+  'all-agents-mcp.json',
 ];
 
-const BROWSER_PROFILES = ['twitter', 'reddit', 'linkedin'];
+const BROWSER_PROFILES = ['reddit', 'linkedin', 'browser-harness'];
 
 function copyDir(src, dest) {
   fs.mkdirSync(dest, { recursive: true });
@@ -71,8 +76,27 @@ function linkOrRelink(target, linkPath) {
   fs.symlinkSync(target, linkPath);
 }
 
+// Locate uv (Astral's Python launcher). The browser-harness MCP server is
+// shebanged through uv so it can pull `mcp` on first run without polluting
+// the system Python. Returns the absolute path if found, or empty string.
+function findUvBin() {
+  const candidates = [
+    path.join(HOME, '.local', 'bin', 'uv'),
+    '/opt/homebrew/bin/uv',
+    '/usr/local/bin/uv',
+    '/usr/bin/uv',
+  ];
+  for (const c of candidates) {
+    if (fs.existsSync(c)) return c;
+  }
+  const which = spawnSync('command', ['-v', 'uv'], { shell: true, encoding: 'utf8' });
+  const found = (which.stdout || '').trim().split('\n')[0];
+  return found && fs.existsSync(found) ? found : '';
+}
+
 function installBrowserAgentConfigs() {
   const nodeBin = path.dirname(process.execPath);
+  const uvBin = findUvBin() || path.join(HOME, '.local', 'bin', 'uv');
   const srcDir = path.join(PKG_ROOT, 'browser-agent-configs');
   const destDir = path.join(HOME, '.claude', 'browser-agent-configs');
   fs.mkdirSync(destDir, { recursive: true });
@@ -90,7 +114,8 @@ function installBrowserAgentConfigs() {
     const tpl = fs.readFileSync(src, 'utf8');
     const out = tpl
       .replace(/__HOME__/g, HOME)
-      .replace(/__NODE_BIN__/g, nodeBin);
+      .replace(/__NODE_BIN__/g, nodeBin)
+      .replace(/__UV_BIN__/g, uvBin);
     fs.writeFileSync(dest, out);
     installed++;
   }
@@ -105,16 +130,92 @@ function installBrowserAgentConfigs() {
   console.log(`  browser profile dirs ready -> ${profilesDir}/{${BROWSER_PROFILES.join(',')}}`);
 }
 
+// Provision the browser-harness toolchain that backs the twitter-harness MCP:
+//   1. install uv (Astral) if missing
+//   2. git-clone browser-use/browser-harness
+//   3. uv tool install -e . (provides the `browser-harness` CLI)
+//   4. ensure `mcp` Python package is importable for server.py
+//   5. copy our shipped server.py into ~/.claude/mcp-servers/browser-harness/
+// All steps are idempotent.
+function installBrowserHarness() {
+  console.log('  setting up browser-harness (twitter-harness MCP backend)...');
+
+  // Step 1: uv. Try the official installer first; fall back to pip.
+  let uvBin = findUvBin();
+  if (!uvBin) {
+    console.log('    uv not found -> installing via Astral installer');
+    const sh = spawnSync('bash', ['-lc', 'curl -LsSf https://astral.sh/uv/install.sh | sh'], { stdio: 'inherit' });
+    if (sh.status !== 0) {
+      console.log('    Astral installer failed; falling back to pip3 install uv');
+      let pip = spawnSync('pip3', ['install', '-q', 'uv'], { stdio: 'inherit' });
+      if (pip.status !== 0) {
+        pip = spawnSync('pip3', ['install', '-q', 'uv', '--break-system-packages'], { stdio: 'inherit' });
+      }
+    }
+    uvBin = findUvBin();
+  }
+  if (!uvBin) {
+    console.warn('  WARNING: uv install failed; twitter-harness MCP server.py will not start.');
+    console.warn('    Install manually: curl -LsSf https://astral.sh/uv/install.sh | sh');
+  } else {
+    console.log(`    uv -> ${uvBin}`);
+  }
+
+  // Step 2 + 3: clone + `uv tool install -e .` browser-harness.
+  const harnessDir = path.join(HOME, 'Developer', 'browser-harness');
+  if (!fs.existsSync(harnessDir)) {
+    fs.mkdirSync(path.dirname(harnessDir), { recursive: true });
+    console.log('    cloning browser-harness from GitHub...');
+    const clone = spawnSync('git', ['clone', '--depth', '1', 'https://github.com/browser-use/browser-harness', harnessDir], { stdio: 'inherit' });
+    if (clone.status !== 0) {
+      console.warn('    WARNING: git clone failed; twitter-harness will not work until you clone manually.');
+    }
+  } else {
+    console.log(`    browser-harness clone exists -> ${harnessDir}`);
+  }
+
+  if (uvBin && fs.existsSync(harnessDir)) {
+    console.log('    installing browser-harness CLI via uv tool...');
+    const install = spawnSync(uvBin, ['tool', 'install', '-e', harnessDir], { stdio: 'inherit' });
+    if (install.status !== 0) {
+      console.warn('    WARNING: `uv tool install -e .` failed; check the output above.');
+    }
+  }
+
+  // Step 4: ensure mcp Python package available (server.py uses `from mcp.server.fastmcp ...`).
+  // server.py is shebanged through `uv run --with mcp ...` so this is belt-and-suspenders;
+  // we install it into the system Python too so a plain `python3 server.py` also works.
+  console.log('    ensuring mcp>=1.0.0 Python package is importable...');
+  let pip = spawnSync('pip3', ['install', '-q', 'mcp>=1.0.0'], { stdio: 'inherit' });
+  if (pip.status !== 0) {
+    pip = spawnSync('pip3', ['install', '-q', 'mcp>=1.0.0', '--break-system-packages'], { stdio: 'inherit' });
+  }
+
+  // Step 5: copy our shipped server.py into the canonical install location.
+  const srcServer = path.join(PKG_ROOT, 'mcp-servers', 'browser-harness', 'server.py');
+  const destServer = path.join(HOME, '.claude', 'mcp-servers', 'browser-harness', 'server.py');
+  if (fs.existsSync(srcServer)) {
+    fs.mkdirSync(path.dirname(destServer), { recursive: true });
+    fs.copyFileSync(srcServer, destServer);
+    try { fs.chmodSync(destServer, 0o755); } catch {}
+    console.log(`    server.py -> ${destServer}`);
+  } else {
+    console.warn(`    WARNING: package missing mcp-servers/browser-harness/server.py (${srcServer})`);
+  }
+}
+
 // Register the three browser-agent MCP servers with Claude so they show up
 // under user scope (writes to ~/.claude.json). Idempotent: parses the output
 // of `claude mcp list` and only calls `add-json` for missing entries.
 // If the `claude` CLI is not on PATH, prints manual instructions and returns.
 function registerBrowserAgentMcpServers() {
   const configDir = path.join(HOME, '.claude', 'browser-agent-configs');
+  // twitter-agent retired 2026-05-19, replaced by twitter-harness (CDP-driven
+  // real Chrome on port 9555 via the browser-harness MCP server).
   const servers = [
-    { name: 'twitter-agent', file: path.join(configDir, 'twitter-agent-mcp.json') },
     { name: 'reddit-agent', file: path.join(configDir, 'reddit-agent-mcp.json') },
     { name: 'linkedin-agent', file: path.join(configDir, 'linkedin-agent-mcp.json') },
+    { name: 'twitter-harness', file: path.join(configDir, 'twitter-harness-mcp.json') },
   ];
 
   const claudeBin = spawnSync('claude', ['--version'], { stdio: 'pipe' });
@@ -138,7 +239,7 @@ function registerBrowserAgentMcpServers() {
       continue;
     }
     // `claude mcp list` prints one server per line starting with the name.
-    // Use a word-boundary check so twitter-agent does not false-match reddit-agent.
+    // Use a word-boundary check so e.g. reddit-agent does not false-match linkedin-agent.
     const re = new RegExp(`(^|\\s)${s.name.replace(/[.*+?^${}()|[\\]\\\\]/g, '\\\\$&')}(:|\\s|$)`, 'm');
     if (re.test(existing)) {
       skipped++;
@@ -288,6 +389,9 @@ function init() {
     installSystemdUnits();
   }
 
+  // Provision the browser-harness toolchain BEFORE writing harness configs so
+  // findUvBin() picks up a freshly-installed uv on first run.
+  installBrowserHarness();
   // Install browser agent MCP configs + profile dirs (skips existing files)
   installBrowserAgentConfigs();
   // Register those MCP servers with Claude so they show up in `claude mcp list`.
@@ -369,6 +473,9 @@ function update() {
     installSystemdUnits();
   }
 
+  // Provision browser-harness (uv + clone + uv tool install + mcp pkg + server.py).
+  // Idempotent: skips steps that are already done.
+  installBrowserHarness();
   // Top up browser agent configs (won't overwrite user customizations)
   installBrowserAgentConfigs();
   // Register any newly added MCP servers with Claude (idempotent).

package/bin/server.js

@@ -5727,16 +5727,24 @@ async function handleApi(req, res) {
         "COALESCE(comments_count, 0)::int AS comments_count, " +
         "CASE WHEN LOWER(posts.platform) IN ('moltbook', 'github', 'github_issues') " +
           "THEN NULL ELSE COALESCE(views, 0)::int END AS views, " +
-        // Score weights comments and upvotes equally (5 each); views are 1/100.
-        // Reddit bakes the OP's self-upvote into the API's `score` field, and our
-        // moltbook_post.py self_upvote() call does the same for Moltbook, so a fresh
-        // post on either platform shows upvotes=1; discount 1, clamped at 0 so
-        // downvoted posts don't go negative.
-        "(COALESCE(comments_count,0) * 5 " +
+        // 2026-05-19: per-post score now matches the unified composite used
+        // by top_performers.SCORE_SQL, the engagement_styles.py picker, and
+        // the dashboard "Posts by Engagement Style" section:
+        //   score = clicks×10 + comments×3 + upvotes_discounted
+        // Click weight ×10 because a real human click outvalues 10 likes of
+        // vibes when ranking engagement quality. Views are deliberately
+        // excluded (viral-by-algorithm ≠ a pattern worth imitating).
+        // Reddit/Moltbook subtract 1 from upvotes to strip the OP self-upvote.
+        // Clicks come from `pl.real_clicks` = the bot-filtered per-hit log
+        // (post_link_clicks). Posts predating the 2026-05-07 log start get 0
+        // clicks here even if they accumulated PostHog $pageview backfill —
+        // by design: the picker uses the same source so the score the model
+        // sees and the score the dashboard shows agree on the same ground truth.
+        "(COALESCE(pl.real_clicks, 0) * 10 " +
+          "+ COALESCE(comments_count,0) * 3 " +
           "+ CASE WHEN LOWER(posts.platform) IN ('reddit', 'moltbook') " +
-            "THEN GREATEST(0, COALESCE(upvotes,0) - 1) * 5 " +
-            "ELSE COALESCE(upvotes,0) * 5 END " +
-          "+ COALESCE(views,0) / 100)::int AS score, " +
+            "THEN GREATEST(0, COALESCE(upvotes,0) - 1) " +
+            "ELSE COALESCE(upvotes,0) END)::int AS score, " +
         "(our_url IS NOT NULL AND thread_url = our_url AND (thread_author IS NULL OR thread_author = our_account)) AS is_thread, " +
         "posted_at, engagement_updated_at, our_content, our_url, thread_url, thread_title, " +
         "LEFT(COALESCE(thread_content, ''), 400) AS thread_content, " +
@@ -5789,10 +5797,13 @@ async function handleApi(req, res) {
         "GREATEST(0, COALESCE(r2.upvotes, 0) - 1)::int AS upvotes, " +
         "COALESCE(r2.comments_count, 0)::int AS comments_count, " +
         "COALESCE(r2.views, 0)::int AS views, " +
-        // Same score formula. Views (Reddit only) contribute /100.
-        "(COALESCE(r2.comments_count,0) * 5 " +
-          "+ GREATEST(0, COALESCE(r2.upvotes,0) - 1) * 5 " +
-          "+ COALESCE(r2.views,0) / 100)::int AS score, " +
+        // 2026-05-19: aligned with the unified composite (see posts branch
+        // for full rationale). Replies have no associated `post_links` row
+        // (we don't mint short links for reply-to-comment threads), so the
+        // clicks term is implicitly 0 here. Score = comments×3 + upvotes_net.
+        // Reddit-only branch → always discount the self-upvote.
+        "(COALESCE(r2.comments_count,0) * 3 " +
+          "+ GREATEST(0, COALESCE(r2.upvotes,0) - 1))::int AS score, " +
         "FALSE AS is_thread, " +
         "r2.replied_at AS posted_at, " +
         "r2.engagement_updated_at, " +
@@ -6138,6 +6149,111 @@ async function handleApi(req, res) {
     return;
   }
 
+  // GET /api/funnel/stats/by_subreddit - Reddit-only post-engagement aggregated
+  // per subreddit. Mirrors the post-engagement columns of /api/funnel/stats
+  // (Posts, Upvotes, Comments, Views, Post Clicks) but the row identity is the
+  // subreddit slug (extracted from posts.thread_url) instead of project_name.
+  // Funnel cells (pageviews/CTAs/bookings) are NOT included because they are
+  // project-level signals from PostHog/Bookings DBs and can't be attributed to
+  // a subreddit. The `projects` array per row tells which projects posted
+  // there in the window so admins can see overlap.
+  if (p === '/api/funnel/stats/by_subreddit' && req.method === 'GET') {
+    const url = new URL(req.url, 'http://localhost');
+    const days = Math.max(1, Math.min(90, parseInt(url.searchParams.get('days') || '7', 10) || 7));
+    const pc = auth.projectClause(req.user, 'project_name', url.searchParams.get('project'));
+    if (!pc.ok) return json(res, { days, subreddits: [] });
+    const pcClause = pc.clause || '';
+    const q = "WITH reddit_posts AS (" +
+                "SELECT " +
+                  "LOWER(SUBSTRING(thread_url FROM 'reddit\\.com/r/([^/]+)/')) AS subreddit, " +
+                  "project_name, id, " +
+                  // Reddit's own UI subtracts the auto-self-upvote; mirror what
+                  // /api/funnel/stats does for the Upvotes column.
+                  "GREATEST(0, COALESCE(upvotes, 0) - 1) AS upvotes, " +
+                  "COALESCE(comments_count, 0) AS comments, " +
+                  "COALESCE(views, 0) AS views " +
+                "FROM posts " +
+                "WHERE LOWER(platform) = 'reddit' " +
+                  "AND posted_at >= NOW() - INTERVAL '" + days + " days' " +
+                  "AND project_name IS NOT NULL " +
+                  "AND our_content <> '(mention - no original post)'" +
+                  pcClause +
+              ") " +
+              "SELECT " +
+                "rp.subreddit, " +
+                "array_agg(DISTINCT rp.project_name) AS projects, " +
+                "COUNT(*)::int AS posts, " +
+                "COALESCE(SUM(rp.upvotes), 0)::int AS upvotes, " +
+                "COALESCE(SUM(rp.comments), 0)::int AS comments, " +
+                "COALESCE(SUM(rp.views), 0)::int AS views, " +
+                "COALESCE(SUM(pl.total_clicks), 0)::int AS post_clicks " +
+              "FROM reddit_posts rp " +
+              "LEFT JOIN (" +
+                "SELECT post_id, SUM(clicks)::int AS total_clicks " +
+                "FROM post_links WHERE post_id IS NOT NULL GROUP BY post_id" +
+              ") pl ON pl.post_id = rp.id " +
+              "WHERE rp.subreddit IS NOT NULL AND rp.subreddit <> '' " +
+              "GROUP BY rp.subreddit " +
+              "ORDER BY upvotes DESC NULLS LAST, posts DESC " +
+              "LIMIT 500";
+    return (async () => {
+      try {
+        const rows = await pq(q);
+        if (rows === null) return json(res, { error: 'db_unavailable' }, 500);
+        // Load subreddit_bans.comment_blocked from config.json and build a
+        // sub→[accounts] map so each row can surface which accounts are
+        // banned from that sub. We're moving toward account-agnostic posting
+        // (different machines can post the same project as different
+        // accounts), so a single sub may have multiple banned accounts over
+        // time. Each comment_blocked entry is {sub, added_at, reason,
+        // account, noticed_by_project} per the 2026-05-11 audit shape.
+        const bansBySub = new Map();
+        try {
+          const cfg = JSON.parse(fs.readFileSync(CONFIG_FILE, 'utf8'));
+          const bans = (cfg && cfg.subreddit_bans) || {};
+          const list = Array.isArray(bans.comment_blocked) ? bans.comment_blocked : [];
+          for (const entry of list) {
+            if (!entry) continue;
+            let slug = null, account = null;
+            if (typeof entry === 'string') {
+              slug = entry;
+            } else if (typeof entry === 'object') {
+              slug = entry.sub || null;
+              account = entry.account || null;
+            }
+            if (!slug) continue;
+            const key = String(slug).toLowerCase();
+            if (!bansBySub.has(key)) bansBySub.set(key, new Set());
+            if (account) bansBySub.get(key).add(String(account));
+          }
+        } catch {}
+        const subreddits = rows.map(r => {
+          const slug = String(r.subreddit || '').toLowerCase();
+          const accounts = bansBySub.has(slug)
+            ? Array.from(bansBySub.get(slug)).sort()
+            : [];
+          return {
+            name: r.subreddit,
+            projects: r.projects || [],
+            banned_accounts: accounts,
+            posts: Number(r.posts) || 0,
+            upvotes: Number(r.upvotes) || 0,
+            comments: Number(r.comments) || 0,
+            views: Number(r.views) || 0,
+            post_clicks: Number(r.post_clicks) || 0,
+          };
+        });
+        return json(res, {
+          generated_at: new Date().toISOString(),
+          days,
+          subreddits,
+        });
+      } catch (e) {
+        return json(res, { error: String(e && e.message || e) }, 500);
+      }
+    })();
+  }
+
   // GET /api/dm/stats - per-project DM funnel (outreach, replies, interest tiers,
   // qualification, bookings, conversions). Window is "active in last N days"
   // (COALESCE(last_message_at, discovered_at)) to match /api/top/dms semantics.
@@ -7792,7 +7908,7 @@ const HTML = `<!DOCTYPE html>
       <div class="style-stats-empty">Loading\u2026</div>
     </div>
   </details>
-  <details class="style-stats-section" id="search-queries-stats" open>
+  <details class="style-stats-section" id="search-queries-stats">
     <summary>
       <span class="style-stats-title"><span class="style-stats-caret">\u25B6</span><span id="search-queries-stats-heading">Search Queries (last 24 hours)</span><span class="stat-card-info" data-tooltip="Per-query stats from twitter_search_attempts + linkedin_search_attempts + reddit_search_attempts. GitHub isn\u2019t represented. Reddit rows show seed concepts (topic buckets) rather than literal search phrases. attempts = times the query was drafted and run. attempts = times the query was drafted and run. candidates_found = sum of tweets/posts the search returned. dud_rate = % of attempts that returned 0. posts_made = candidates from this query that we actually posted to. avg_engagement = comments\u00D73 + upvotes on those resulting posts (same formula as top_performers.py). Honors Window/Platform/Project filters above.">i</span></span>
       <span class="style-stats-total" id="search-queries-stats-total"></span>
@@ -7808,6 +7924,15 @@ const HTML = `<!DOCTYPE html>
       <div class="style-stats-empty">Loading\u2026</div>
     </div>
   </details>
+  <details class="style-stats-section" id="subreddit-stats" open>
+    <summary>
+      <span class="style-stats-title"><span class="style-stats-caret">\u25B6</span><span id="subreddit-stats-heading">Per-Subreddit Stats (last 24 hours)</span><span class="stat-card-info" data-tooltip="Reddit posts only, grouped by subreddit. Same post-engagement columns as Project Funnel Stats above, but the row identity is the subreddit slug (extracted from posts.thread_url) instead of project_name. Funnel cells (pageviews / CTAs / bookings) are NOT shown here because they\u2019re project-level signals and can\u2019t be attributed to a subreddit. The Projects column lists which projects posted to that sub in the window.">i</span></span>
+      <span class="style-stats-total" id="subreddit-stats-total"></span>
+    </summary>
+    <div id="subreddit-stats-body">
+      <div class="style-stats-empty">Loading\u2026</div>
+    </div>
+  </details>
 </div>
 
 <div class="content hidden" id="tab-trends">
@@ -10783,6 +10908,8 @@ async function reloadStatsTabSections() {
   if (dmEl && dmEl.open) pending.push(loadDmStats(true));
   const sqEl = document.getElementById('search-queries-stats');
   if (sqEl && sqEl.open) pending.push(loadSearchQueriesStats(true));
+  const subEl = document.getElementById('subreddit-stats');
+  if (subEl && subEl.open) pending.push(loadSubredditStats(true));
   try { await Promise.allSettled(pending); }
   finally { document.body.classList.remove('sa-stats-busy'); }
 }
@@ -10801,6 +10928,8 @@ function syncStatsHeadings() {
   if (dm) dm.textContent = 'DM Funnel Stats (' + win.labelLong + ')';
   const sq = document.getElementById('search-queries-stats-heading');
   if (sq) sq.textContent = 'Search Queries (' + win.labelLong + ')';
+  const sub = document.getElementById('subreddit-stats-heading');
+  if (sub) sub.textContent = 'Per-Subreddit Stats (' + win.labelLong + ')';
 }
 function syncStatusHeadings() {
   const win = currentStatusWindow();
@@ -13204,6 +13333,87 @@ function renderFunnelStats(payload) {
     '</div>');
 }
 
+// Per-subreddit post-engagement stats. Reddit-only sibling of renderFunnelStats:
+// same Posts / Upvotes / Comments / Views / Post Clicks columns, but the row
+// identity is the subreddit slug. No funnel cells (pageviews / CTAs / bookings)
+// because those are project-level and can\u2019t be attributed to a subreddit.
+let _subredditStatsTableState = { sortField: 'upvotes', sortDir: 'desc', filters: {} };
+function renderSubredditStats(payload) {
+  const body = document.getElementById('subreddit-stats-body');
+  const totalEl = document.getElementById('subreddit-stats-total');
+  if (!body) return;
+  if (payload && payload.error) {
+    if (totalEl) totalEl.textContent = 'error';
+    body.innerHTML = '<div class="style-stats-empty">' + escapeHtml(payload.error) + '</div>';
+    return;
+  }
+  const rows = (payload && payload.subreddits) || [];
+  if (!rows.length) {
+    if (totalEl) totalEl.textContent = '0 subreddits';
+    body.innerHTML = '<div class="style-stats-empty">No Reddit posts in this window.</div>';
+    return;
+  }
+  if (totalEl) totalEl.textContent = rows.length + ' subreddit' + (rows.length === 1 ? '' : 's');
+  const fmt = n => (Number(n) || 0).toLocaleString();
+  const normalized = rows.map(r => ({
+    name:         r.name || '',
+    projects:     Array.isArray(r.projects) ? r.projects.join(', ') : '',
+    banned:       Array.isArray(r.banned_accounts) ? r.banned_accounts.join(', ') : '',
+    posts:        Number(r.posts) || 0,
+    upvotes:      Number(r.upvotes) || 0,
+    comments:     Number(r.comments) || 0,
+    views:        Number(r.views) || 0,
+    post_clicks:  Number(r.post_clicks) || 0,
+  }));
+  const fmtSubName = v => {
+    const slug = String(v || '').replace(/[^a-zA-Z0-9_\-]/g, '');
+    if (!slug) return escapeHtml(v);
+    return '<a href="https://www.reddit.com/r/' + encodeURIComponent(slug) + '/" target="_blank" rel="noopener">r/' + escapeHtml(slug) + '</a>';
+  };
+  const fmtProjects = v => {
+    const s = String(v || '');
+    if (!s) return '\u2014';
+    return '<span style="color:var(--text-muted);font-size:12px;" title="' + escapeHtml(s) + '">' + escapeHtml(s) + '</span>';
+  };
+  // Banned column: comma-separated list of Reddit handles whose
+  // subreddit_bans.comment_blocked entry matches this sub. Each handle
+  // links to the account's Reddit profile so an admin can verify the ban
+  // is real (e.g. confirm shadow-reject vs hard ban). Account-agnostic:
+  // multiple machines may post the same project under different accounts,
+  // so a sub can carry several bans (one per offending account).
+  const fmtBannedAccounts = v => {
+    const s = String(v || '').trim();
+    if (!s) return '<span style="color:var(--text-muted);">\u2014</span>';
+    const handles = s.split(',').map(x => x.trim()).filter(Boolean);
+    const parts = handles.map(h => {
+      const safe = h.replace(/[^a-zA-Z0-9_\-]/g, '');
+      if (!safe) return escapeHtml(h);
+      return '<a href="https://www.reddit.com/user/' + encodeURIComponent(safe) + '/" target="_blank" rel="noopener" style="color:var(--danger,#e74c3c);">u/' + escapeHtml(safe) + '</a>';
+    });
+    return parts.join(', ');
+  };
+  mountSortableTable({
+    containerId: 'subreddit-stats-body',
+    state: _subredditStatsTableState,
+    storageKey: 'sa.subredditStatsTable.v1',
+    rows: normalized,
+    showTotals: true,
+    columns: [
+      { key: 'name',        label: 'Subreddit',   type: 'text',    align: 'left',  formatter: fmtSubName },
+      { key: 'projects',    label: 'Projects',    type: 'text',    align: 'left',  formatter: fmtProjects },
+      { key: 'banned',      label: 'Banned',      type: 'text',    align: 'left',  formatter: fmtBannedAccounts,
+        helpText: 'Reddit accounts whose subreddit_bans.comment_blocked entry matches this sub. Multiple handles means multiple agent accounts have been banned from posting here. "\u2014" means no recorded ban for this sub. Source: config.json subreddit_bans.comment_blocked.' },
+      { key: 'posts',       label: 'Posts',       type: 'numeric', align: 'right', formatter: fmt },
+      { key: 'upvotes',     label: 'Upvotes',     type: 'numeric', align: 'right', formatter: fmt,
+        helpText: 'Sum of upvotes on Reddit posts created in the selected window, minus the auto-self-upvote per post (matches the Project Funnel Stats column).' },
+      { key: 'comments',    label: 'Comments',    type: 'numeric', align: 'right', formatter: fmt },
+      { key: 'views',       label: 'Views',       type: 'numeric', align: 'right', formatter: fmt },
+      { key: 'post_clicks', label: 'Post Clicks', type: 'numeric', align: 'right', formatter: fmt,
+        helpText: 'Clicks on /r/<code> short links minted for posts created in the selected window (no period-total here \u2014 this is a fast Reddit-only path).' },
+    ],
+  });
+}
+
 let _dmStatsTableState = { sortField: 'sent', sortDir: 'desc', filters: {} };
 function renderDmStats(payload) {
   const body = document.getElementById('dm-stats-body');
@@ -16392,6 +16602,34 @@ async function loadFunnelStats(force) {
   }
 }
 
+let _subredditStatsLoadedFor = null;
+let _subredditStatsLoading = false;
+async function loadSubredditStats(force) {
+  if (_subredditStatsLoading) return;
+  if (saAuthNotReady()) return;
+  const days = currentStatsWindow().days;
+  const proj = currentStatsProject();
+  const loadKey = days + '|' + proj;
+  if (_subredditStatsLoadedFor === loadKey && !force) return;
+  _subredditStatsLoading = true;
+  const totalEl = document.getElementById('subreddit-stats-total');
+  const body = document.getElementById('subreddit-stats-body');
+  if (totalEl) totalEl.textContent = 'loading\u2026';
+  if (body) body.innerHTML = '<div class="style-stats-empty">Loading\u2026</div>';
+  try {
+    const params = ['days=' + days];
+    if (proj && proj !== 'all') params.push('project=' + encodeURIComponent(proj));
+    const res = await fetch('/api/funnel/stats/by_subreddit?' + params.join('&'));
+    const data = await res.json();
+    renderSubredditStats(data);
+    _subredditStatsLoadedFor = loadKey;
+  } catch (e) {
+    if (body) body.innerHTML = '<div class="style-stats-empty">Failed to load.</div>';
+  } finally {
+    _subredditStatsLoading = false;
+  }
+}
+
 let _dmStatsLoadedFor = null;
 let _dmStatsLoading = false;
 async function loadDmStats(force) {
@@ -16707,6 +16945,8 @@ _saInstallDeleteListener();
     if (funnelEl && funnelEl.open) loadFunnelStats(true);
     const dmEl = document.getElementById('dm-stats');
     if (dmEl && dmEl.open) loadDmStats(true);
+    const subEl = document.getElementById('subreddit-stats');
+    if (subEl && subEl.open) loadSubredditStats(true);
   });
   syncStatsHeadings();
 })();
@@ -16817,6 +17057,15 @@ _saInstallDeleteListener();
   }
 })();
 
+(function wireSubredditStats() {
+  const el = document.getElementById('subreddit-stats');
+  if (!el) return;
+  el.addEventListener('toggle', () => {
+    try { window.posthog && window.posthog.capture('section_toggle', { section: 'subreddit-stats', open: !!el.open }); } catch (er) {}
+    if (el.open) loadSubredditStats();
+  });
+})();
+
 (function wireCostStats() {
   const el = document.getElementById('cost-stats');
   if (!el) return;
@@ -16898,6 +17147,8 @@ function saStartApp() {
   if (dmEl && dmEl.open) loadDmStats();
   const sqEl = document.getElementById('search-queries-stats');
   if (sqEl && sqEl.open) loadSearchQueriesStats();
+  const subEl = document.getElementById('subreddit-stats');
+  if (subEl && subEl.open) loadSubredditStats();
   setInterval(loadActivityStats, 300000);
   setInterval(loadCohortStats, 300000);
   setInterval(loadStyleStats, 300000);

package/browser-agent-configs/all-agents-mcp.json

@@ -0,0 +1,49 @@
+{
+  "mcpServers": {
+    "reddit-agent": {
+      "type": "stdio",
+      "command": "/usr/bin/python3",
+      "args": [
+        "__HOME__/social-autoposter/scripts/mcp_lock_proxy.py",
+        "--lock-name", "reddit-browser",
+        "--ttl", "90",
+        "--",
+        "npx",
+        "@playwright/mcp@latest",
+        "--config",
+        "__HOME__/.claude/browser-agent-configs/reddit-agent.json"
+      ],
+      "env": {
+        "PATH": "__NODE_BIN__:/usr/bin:/bin",
+        "BROWSER_LOCK_NAME": "reddit-browser",
+        "BROWSER_LOCK_TTL": "90"
+      }
+    },
+    "linkedin-agent": {
+      "type": "stdio",
+      "command": "npx",
+      "args": [
+        "@playwright/mcp@latest",
+        "--config",
+        "__HOME__/.claude/browser-agent-configs/linkedin-agent.json"
+      ],
+      "env": {
+        "PATH": "__NODE_BIN__:/usr/bin:/bin"
+      }
+    },
+    "twitter-harness": {
+      "type": "stdio",
+      "command": "__UV_BIN__",
+      "args": [
+        "run",
+        "--quiet",
+        "--with",
+        "mcp",
+        "__HOME__/.claude/mcp-servers/browser-harness/server.py"
+      ],
+      "env": {
+        "PATH": "__HOME__/.local/bin:__NODE_BIN__:/opt/homebrew/bin:/usr/local/bin:/usr/bin:/bin"
+      }
+    }
+  }
+}

package/browser-agent-configs/twitter-harness-mcp.json

@@ -0,0 +1,18 @@
+{
+  "mcpServers": {
+    "twitter-harness": {
+      "type": "stdio",
+      "command": "__UV_BIN__",
+      "args": [
+        "run",
+        "--quiet",
+        "--with",
+        "mcp",
+        "__HOME__/.claude/mcp-servers/browser-harness/server.py"
+      ],
+      "env": {
+        "PATH": "__HOME__/.local/bin:__NODE_BIN__:/opt/homebrew/bin:/usr/local/bin:/usr/bin:/bin"
+      }
+    }
+  }
+}