social-autoposter 1.3.0 → 1.3.2

package/bin/cli.js

@@ -18,6 +18,7 @@ const COPY_TARGETS = [
   'scripts',
   'schema-postgres.sql',
   'config.example.json',
+  'requirements.txt',
   'SKILL.md',
   'skill',
   'setup',
@@ -403,14 +404,21 @@ function update() {
 // browser binary; we run `playwright install chromium` after the pip install.
 function installPythonDeps() {
   const reqPath = path.join(PKG_ROOT, 'requirements.txt');
-  const args = fs.existsSync(reqPath)
+  const base = fs.existsSync(reqPath)
     ? ['install', '-r', reqPath, '-q']
     : ['install', '-q', 'psycopg2-binary', 'playwright'];
   console.log('  installing Python deps (psycopg2-binary, playwright, ...)');
-  const r = spawnSync('pip3', args, { stdio: 'inherit' });
+  // Debian/Ubuntu 23+ ship a PEP 668 marker that blocks pip3 against the
+  // system Python without --break-system-packages. Try without first
+  // (safer on macOS) and retry with the flag if the marker fires.
+  let r = spawnSync('pip3', base, { stdio: 'inherit' });
+  if (r.status !== 0) {
+    console.log('  retrying with --break-system-packages (PEP 668 environments)');
+    r = spawnSync('pip3', [...base, '--break-system-packages'], { stdio: 'inherit' });
+  }
   if (r.status !== 0) {
     console.warn('  WARNING: pip3 install failed — run manually:');
-    console.warn(`    pip3 ${args.join(' ')}`);
+    console.warn(`    pip3 ${base.join(' ')} --break-system-packages`);
     return;
   }
   // Playwright needs its browser binary downloaded separately. Chromium

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "social-autoposter",
-  "version": "1.3.0",
+  "version": "1.3.2",
   "description": "Automated social posting pipeline for Reddit, X/Twitter, LinkedIn, and Moltbook. Install as a Claude Code agent skill.",
   "bin": {
     "social-autoposter": "bin/cli.js"

package/scripts/engagement_styles.py

@@ -399,7 +399,8 @@ def _fetch_style_stats(platform):
         _sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
         from http_api import api_get
         resp = api_get("/api/v1/engagement-styles/style-stats", {"platform": platform})
-        stats = (resp or {}).get("stats") or {}
+        data = (resp or {}).get("data") or {}
+        stats = data.get("stats") or {}
         return {
             name: {"n": int(v.get("n", 0)), "avg_up": float(v.get("avg_up", 0.0))}
             for name, v in stats.items()
@@ -485,7 +486,8 @@ def _last_picks(platform, limit=10):
             "/api/v1/engagement-styles/last-picks",
             {"platform": platform, "limit": int(limit)},
         )
-        picks = (resp or {}).get("picks") or []
+        data = (resp or {}).get("data") or {}
+        picks = data.get("picks") or []
         return [str(p) for p in picks if p]
     except Exception:
         return []

package/scripts/twitter_browser.py

@@ -34,10 +34,6 @@ import time
 
 
 PROFILE_DIR = os.path.expanduser("~/.claude/browser-profiles/twitter")
-# Camoufox (Firefox-stealth) parallel profile. Used when TWITTER_ENGINE=camoufox.
-# Seeded from Chrome cookies via scripts/dump_twitter_storage.py on first use.
-CAMOUFOX_PROFILE_DIR = os.path.expanduser("~/.camoufox-mcp/profiles/twitter")
-CAMOUFOX_STORAGE_JSON = os.path.expanduser("~/.camoufox-mcp/storage/twitter.storage.json")
 LOCK_FILE = os.path.expanduser("~/.claude/twitter-agent-lock.json")
 LOCK_EXPIRY = 300  # Must match twitter-agent-lock.sh
 LOCK_WAIT_MAX = 45  # seconds to wait for lock to free before giving up
@@ -270,91 +266,15 @@ def _refresh_browser_lock():
         pass
 
 
-def _ensure_camoufox_seeded():
-    """Make sure ~/.camoufox-mcp/storage/twitter.storage.json exists with valid
-    cookies. If not, dump them from the live Chrome twitter-agent profile via
-    scripts/dump_twitter_storage.py. One-shot; cheap if already seeded.
-    """
-    if os.path.exists(CAMOUFOX_STORAGE_JSON):
-        try:
-            with open(CAMOUFOX_STORAGE_JSON) as f:
-                data = json.load(f)
-            has_auth = any(c.get("name") == "auth_token" for c in data.get("cookies", []))
-            if has_auth:
-                return
-        except Exception:
-            pass
-    dumper = os.path.join(os.path.dirname(os.path.abspath(__file__)),
-                          "dump_twitter_storage.py")
-    if os.path.exists(dumper):
-        subprocess.run([sys.executable, dumper], check=False, timeout=60)
-
-
-def _launch_camoufox_browser_and_page(playwright):
-    """Camoufox path: Firefox-engine browser with stealth fingerprint patches.
-
-    Headless-safe (camoufox patches Firefox at the C++ layer so headless looks
-    identical to headed), so it sidesteps Twitter's 2026-04-30 headless-Chrome
-    DM strip-down. Uses its own persistent profile dir to avoid colliding with
-    the Chrome twitter-agent's profile lock.
-
-    Returns (context, page, False) to match the existing non-CDP contract.
-    """
-    from camoufox.sync_api import NewBrowser
-    os.makedirs(CAMOUFOX_PROFILE_DIR, exist_ok=True)
-    _ensure_camoufox_seeded()
-    headless_env = os.environ.get("TWITTER_CAMOUFOX_HEADLESS", "true").lower()
-    headless = headless_env not in ("0", "false", "no")
-    context = NewBrowser(
-        playwright,
-        persistent_context=True,
-        user_data_dir=CAMOUFOX_PROFILE_DIR,
-        headless=headless,
-        humanize=True,
-        block_webrtc=True,
-        geoip=True,
-    )
-    # First-run cookie hydration from the seeded storage JSON (only if the
-    # persistent profile doesn't already carry an auth_token).
-    try:
-        existing = context.cookies()
-        if not any(c.get("name") == "auth_token" for c in existing):
-            if os.path.exists(CAMOUFOX_STORAGE_JSON):
-                with open(CAMOUFOX_STORAGE_JSON) as f:
-                    state = json.load(f)
-                if state.get("cookies"):
-                    context.add_cookies(state["cookies"])
-    except Exception as e:
-        print(f"[twitter_browser] camoufox cookie hydration failed: {e}",
-              file=sys.stderr)
-    page = context.new_page()
-    return context, page, False
-
-
 def get_browser_and_page(playwright):
     """Connect to the twitter-agent MCP browser via CDP, or launch a new one.
 
     Returns (browser, page, is_cdp). When is_cdp=True, `page` is a reused
     existing Twitter tab (navigate it, don't close it). When is_cdp=False,
     it's a new headless page.
-
-    Set TWITTER_ENGINE=camoufox to swap in a Firefox-stealth engine instead of
-    Chrome. Same return shape; all callers untouched. See
-    _launch_camoufox_browser_and_page() above.
     """
     _acquire_browser_lock()
 
-    if os.environ.get("TWITTER_ENGINE", "").lower() == "camoufox":
-        try:
-            return _launch_camoufox_browser_and_page(playwright)
-        except Exception as e:
-            _release_browser_lock()
-            print(json.dumps({
-                "success": False,
-                "error": f"camoufox launch failed: {e}"
-            }))
-            sys.exit(1)
-
     cdp_port = find_twitter_cdp_port()
 
     if cdp_port:
@@ -583,19 +503,14 @@ def reply_to_tweet(tweet_url, text, apply_campaigns=True):
 
         try:
             # Set up Network interception to capture CreateTweet response.
-            # Two parallel paths so both engines work:
-            #   (a) page.on("response") — engine-agnostic. Works on Chromium
-            #       AND on Firefox/camoufox. This is the only path that fires
-            #       when TWITTER_ENGINE=camoufox (Firefox doesn't support CDP).
-            #   (b) CDP Network.responseReceived — Chromium-only. Slightly
-            #       faster + less body-fetch overhead. Skipped on Firefox
-            #       because new_cdp_session raises "CDP session is only
-            #       available in Chromium" there.
+            # Two parallel paths for redundancy:
+            #   (a) page.on("response") — Playwright's event-loop hook.
+            #   (b) CDP Network.responseReceived — slightly faster + less
+            #       body-fetch overhead, Chromium-only.
             # Both write into _created_tweet_ids; dedup-on-append keeps the
             # list a set of unique rest_ids regardless of which path fired.
             _cdp_session = None
             _created_tweet_ids = []
-            _is_firefox = os.environ.get("TWITTER_ENGINE", "").lower() == "camoufox"
 
             def _on_response_event(resp):
                 # Engine-agnostic CreateTweet capture. Filter by URL FIRST so
@@ -620,35 +535,34 @@ def reply_to_tweet(tweet_url, text, apply_campaigns=True):
 
             page.on("response", _on_response_event)
 
-            if not _is_firefox:
-                try:
-                    _cdp_session = page.context.new_cdp_session(page)
-                    _cdp_session.send("Network.enable")
-
-                    def _on_cdp_response(params):
-                        try:
-                            url = params.get("response", {}).get("url", "")
-                            if "CreateTweet" in url:
-                                body_resp = _cdp_session.send(
-                                    "Network.getResponseBody",
-                                    {"requestId": params["requestId"]},
-                                )
-                                data = json.loads(body_resp.get("body", "{}"))
-                                rest_id = (
-                                    data.get("data", {})
-                                    .get("create_tweet", {})
-                                    .get("tweet_results", {})
-                                    .get("result", {})
-                                    .get("rest_id")
-                                )
-                                if rest_id and rest_id not in _created_tweet_ids:
-                                    _created_tweet_ids.append(rest_id)
-                        except Exception:
-                            pass
+            try:
+                _cdp_session = page.context.new_cdp_session(page)
+                _cdp_session.send("Network.enable")
 
-                    _cdp_session.on("Network.responseReceived", _on_cdp_response)
-                except Exception:
-                    pass
+                def _on_cdp_response(params):
+                    try:
+                        url = params.get("response", {}).get("url", "")
+                        if "CreateTweet" in url:
+                            body_resp = _cdp_session.send(
+                                "Network.getResponseBody",
+                                {"requestId": params["requestId"]},
+                            )
+                            data = json.loads(body_resp.get("body", "{}"))
+                            rest_id = (
+                                data.get("data", {})
+                                .get("create_tweet", {})
+                                .get("tweet_results", {})
+                                .get("result", {})
+                                .get("rest_id")
+                            )
+                            if rest_id and rest_id not in _created_tweet_ids:
+                                _created_tweet_ids.append(rest_id)
+                    except Exception:
+                        pass
+
+                _cdp_session.on("Network.responseReceived", _on_cdp_response)
+            except Exception:
+                pass
 
             page.goto(tweet_url, wait_until="domcontentloaded")
             page.wait_for_timeout(5000)
@@ -718,8 +632,7 @@ def reply_to_tweet(tweet_url, text, apply_campaigns=True):
             # Method 1: CDP network interception (most reliable)
             if _created_tweet_ids:
                 reply_url = f"https://x.com/{OUR_HANDLE}/status/{_created_tweet_ids[-1]}"
-                src = "response-listener" if _is_firefox else "CDP+response-listener"
-                print(f"[reply_url] captured via {src}: {reply_url}", file=sys.stderr)
+                print(f"[reply_url] captured via CDP+response-listener: {reply_url}", file=sys.stderr)
 
             # Method 2: DOM diff (check if new reply links appeared)
             if not reply_url:

package/skill/run-twitter-cycle.sh

@@ -48,47 +48,18 @@ QUERIES_FILE="/tmp/twitter_cycle_queries_$(date +%s).json"
 RUN_START=$(date +%s)
 
 # ----------------------------------------------------------------------------
-# Engine switch: TWITTER_ENGINE controls which browser the Claude SCAN/PREP
-# phases drive AND which engine the python posting path (twitter_browser.py)
-# uses. Defaults to chrome so unset env = today's behavior, zero drift.
+# Browser: Playwright MCP attached to Chrome twitter-agent profile at
+# ~/.claude/browser-profiles/twitter. (Camoufox/Firefox engine was carved out
+# 2026-05-13; only Chrome is supported now.)
 #
-# TWITTER_ENGINE=chrome   (default) Playwright MCP -> Chrome twitter-agent
-#                                   profile at ~/.claude/browser-profiles/twitter
-# TWITTER_ENGINE=camoufox          camoufox MCP -> Firefox-stealth profile at
-#                                   ~/.camoufox-mcp/profiles/twitter
-#
-# All Claude SDK invocations in this script use $TW_MCP_CONFIG so the swap is
-# one variable. Both prompt strings get $TW_ENGINE_PREFIX prepended; for the
-# chrome path the prefix is empty (no behavior change). For camoufox the
-# prefix tells Claude to call camoufox_open(name="twitter") first to attach
-# to the already-logged-in session — the MCP is session-based with no
-# implicit default.
+# Vars are kept (TW_MCP_CONFIG, TW_BROWSER_PROFILE, TW_ENGINE_PREFIX) so the
+# downstream Claude SDK calls and singleton-cleanup hooks need no edits.
+# TW_ENGINE_PREFIX is empty by design; it used to prepend engine-specific
+# instructions to the scan/prep prompts.
 # ----------------------------------------------------------------------------
-TWITTER_ENGINE="${TWITTER_ENGINE:-chrome}"
-if [ "$TWITTER_ENGINE" = "camoufox" ]; then
-    TW_MCP_CONFIG="$HOME/.claude/browser-agent-configs/twitter-agent-camoufox-mcp.json"
-    TW_BROWSER_PROFILE="$HOME/.camoufox-mcp/profiles/twitter"
-    # Export so python subprocesses (twitter_browser.py via twitter_post_plan.py
-    # and any other indirect callers) opt into camoufox via os.environ check.
-    export TWITTER_ENGINE
-    export TWITTER_CAMOUFOX_HEADLESS="${TWITTER_CAMOUFOX_HEADLESS:-true}"
-    # Headless flag for the Claude SDK scan/prep prompts. The camoufox MCP's
-    # camoufox_open() takes `headless` as a tool argument, NOT an env var, so
-    # the prefix must explicitly tell Claude to pass it. Default headless=True;
-    # set TWITTER_CAMOUFOX_HEADLESS=false to debug visually.
-    if [ "${TWITTER_CAMOUFOX_HEADLESS}" = "true" ]; then
-        TW_OPEN_CALL='camoufox_open(name="twitter", headless=True)'
-    else
-        TW_OPEN_CALL='camoufox_open(name="twitter")'
-    fi
-    TW_ENGINE_PREFIX="ENGINE: camoufox (Firefox-stealth via camoufox MCP). Before doing anything else, call ${TW_OPEN_CALL} to attach to the already-logged-in twitter session at ~/.camoufox-mcp/profiles/twitter. Use camoufox_navigate / camoufox_evaluate / camoufox_snapshot / camoufox_click / camoufox_type — all require name=\"twitter\". Do NOT use any browser_* tool; they are not available on this engine.
-
-"
-else
-    TW_MCP_CONFIG="$HOME/.claude/browser-agent-configs/twitter-agent-mcp.json"
-    TW_BROWSER_PROFILE="$HOME/.claude/browser-profiles/twitter"
-    TW_ENGINE_PREFIX=""
-fi
+TW_MCP_CONFIG="$HOME/.claude/browser-agent-configs/twitter-agent-mcp.json"
+TW_BROWSER_PROFILE="$HOME/.claude/browser-profiles/twitter"
+TW_ENGINE_PREFIX=""
 # Tweets older than this are no longer worth replying to. Pending rows older
 # than this are hard-expired by Phase 0; younger pending rows are salvaged
 # from prior cycles into this batch.

package/scripts/dump_twitter_storage.py

@@ -1,25 +0,0 @@
-"""One-shot: launch Playwright against twitter-agent profile, dump storage_state.json, close."""
-import os, sys, json
-from pathlib import Path
-from playwright.sync_api import sync_playwright
-
-PROFILE = str(Path.home() / ".claude/browser-profiles/twitter")
-OUT     = str(Path.home() / ".camoufox-mcp/storage/twitter.storage.json")
-Path(OUT).parent.mkdir(parents=True, exist_ok=True)
-
-with sync_playwright() as p:
-    ctx = p.chromium.launch_persistent_context(
-        PROFILE,
-        headless=True,
-        args=["--no-sandbox"],
-    )
-    state = ctx.storage_state(path=OUT)
-    cookies = state.get("cookies", [])
-    tw = [c for c in cookies if any(d in c.get("domain","") for d in ("twitter.com","x.com"))]
-    print(f"total cookies: {len(cookies)}, twitter/x: {len(tw)}")
-    if tw:
-        names = sorted({c['name'] for c in tw})
-        print("twitter cookie names:", names[:20])
-    ctx.close()
-
-print(f"wrote {OUT}")