@m13v/s4l 1.7.5-rc.30 → 1.7.5-rc.32

package/mcp/dist/version.json

@@ -1,4 +1,4 @@
 {
-  "version": "1.7.5-rc.30",
-  "installedAt": "2026-07-16T00:37:45.238Z"
+  "version": "1.7.5-rc.32",
+  "installedAt": "2026-07-16T01:09:13.414Z"
 }

package/mcp/manifest.json

@@ -2,7 +2,7 @@
   "dxt_version": "0.1",
   "name": "social-autoposter",
   "display_name": "S4L",
-  "version": "1.7.5-rc.30",
+  "version": "1.7.5-rc.32",
   "description": "Draft, review, approve, and autopilot X/Twitter posts.",
   "long_description": "## **⚠️ The disclaimer above is generic Claude boilerplate.** Anthropic shows the same warning on every plugin regardless of what it does; any plugin has the same level of access as any app you download from the internet.\n\nS4L is an open source product developed by Mediar.ai Incorporated, a VC-backed San Francisco-based startup.\n\nTo get started:\n\n1\\. Copy this prompt: **Set me up on S4L plugin end to end**\n\n2\\. Quit with CMD+Q, reopen Claude, paste into a new chat.\n\nWhat happens next:\n\n* About every 5 minutes S4L scans X for posts that match your topics and drafts replies in your voice.\n* Drafts show up as review cards, usually the first within a few minutes. Nothing is posted automatically; you approve each one.\n* Posting autopilot stays off until you explicitly turn it on.",
   "author": {

package/mcp/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@m13v/s4l-mcp",
-  "version": "1.7.5-rc.30",
+  "version": "1.7.5-rc.32",
   "private": true,
   "description": "Desktop MCP client for social-autoposter (X/Twitter rail): manual draft/review/approve loop, autopilot control, and stats. Thin wrapper over the existing pipeline scripts.",
   "license": "MIT",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@m13v/s4l",
-  "version": "1.7.5-rc.30",
+  "version": "1.7.5-rc.32",
   "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/merge_review_queue.py

@@ -107,6 +107,17 @@ def _atomic_write(path: str, obj) -> None:
 
 def _dedup_key(c: dict) -> str:
     """Match submit_drafts: dedup by the thread/candidate URL, else candidate_id."""
+    # Sandbox cards (experiments.sandbox=true) replay real historical threads
+    # ON PURPOSE, including ones already dealt with for real -- that's the
+    # whole point of testing against past data. Namespace their key so they
+    # never collide with (and get silently swallowed by) a real card for the
+    # same URL, and keep the reply text IN the key so two different prompt
+    # variants drafting the same thread both surface as distinct cards
+    # instead of the second dropping as "already seen" (identical reruns
+    # still naturally dedupe against each other, which is fine).
+    if (c.get("experiments") or {}).get("sandbox"):
+        url = c.get("candidate_url") or c.get("tweet_url") or c.get("thread_url") or ""
+        return f"sandbox:{url}:{(c.get('reply_text') or '')[:80]}"
     for k in ("candidate_url", "tweet_url", "thread_url", "candidate_id"):
         v = c.get(k)
         if v:

package/skill/lib/harness-common.sh

@@ -2,6 +2,15 @@
 # harness-common.sh - the shared engine behind every per-platform browser
 # backend (twitter-backend.sh, reddit-backend.sh, linkedin-backend.sh).
 #
+# LOCAL OVERRIDES: ~/.social-autoposter-mcp/harness-local.env is sourced (when
+# present) before any launch decision. It is machine-local (never shipped),
+# for per-box experiments without a release: e.g. pointing one operator box's
+# harness at Chrome Beta (BH_CHROME_BIN=...) or adding launch flags
+# (HC_EXTRA_FLAGS="--no-startup-window") while every other install keeps
+# stock behavior. Values here are plain shell assignments; keep it tiny.
+# shellcheck disable=SC1090
+[ -f "$HOME/.social-autoposter-mcp/harness-local.env" ] && . "$HOME/.social-autoposter-mcp/harness-local.env" 2>/dev/null || true
+#
 # Born 2026-07-14 from the three near-duplicate backends. Every hard-won fix
 # used to land in ONE backend and drift: the CDP wedge detector, the two-strike
 # tolerance, the focus-safe launcher, the occlusion flags, and the repo-relative
@@ -287,6 +296,15 @@ hc_ensure_browser() {
         # 2026-07-14/15 EXC_BREAKPOINT (deliberate internal abort) crash dumps
         # arriving ~once per cycle with zero jetsam/OOM evidence. Playwright
         # and Puppeteer both launch with it for exactly this reason.
+        # With --no-startup-window (focus-steal prevention: Chrome creates NO
+        # window at launch, so macOS has nothing to activate; the first
+        # pipeline op mints its tab via the background Target.createTarget
+        # path) a positional URL argument would force a window and defeat the
+        # flag — drop the launch URL in that mode.
+        local _launch_url="${HC_LAUNCH_URL:-about:blank}"
+        case " ${HC_EXTRA_FLAGS:-} " in
+            *" --no-startup-window "*) _launch_url="" ;;
+        esac
         launch_harness_chrome "$_chrome_bin" "$_prof" \
             --remote-debugging-port="$_port" \
             --user-data-dir="$_prof" \
@@ -296,7 +314,7 @@ hc_ensure_browser() {
             --disable-backgrounding-occluded-windows \
             ${HC_EXTRA_FLAGS:-} \
             ${_os_extra} \
-            "${HC_LAUNCH_URL:-about:blank}"
+            ${_launch_url}
         for _i in 1 2 3 4 5 6 7 8 9 10 11 12; do
             curl -sf --max-time 2 -o /dev/null "${_def_url}/json/version" 2>/dev/null && break
             sleep 1

package/skill/run-reddit-search.sh

@@ -284,8 +284,23 @@ python3 "$REPO_DIR/scripts/reddit_browser_lock.py" acquire --timeout 60 --ttl 30
 # old ensure_browser_healthy "reddit" ps-scan-for-MCP-Chrome path; the harness is
 # the single browser the whole Reddit pipeline now rides (REDDIT_CDP_URL points
 # every child Python proc at it). Falls back to ensure_browser_healthy on failure.
-if ! ensure_reddit_browser_for_backend 2>&1 | tee -a "$LOG_FILE"; then
-    log "WARNING: reddit-harness bootstrap failed; falling back to ensure_browser_healthy reddit"
+ensure_reddit_browser_for_backend 2>&1 | tee -a "$LOG_FILE" || true
+_ensure_rc="${PIPESTATUS[0]}"
+if [ "$_ensure_rc" = "78" ]; then
+    # Reserved skip code (harness-common.sh:182-185): the pre-launch hook
+    # deliberately deferred because a poster is mid-drain on the ONE shared
+    # tab. This is NOT a bootstrap failure — falling back to
+    # ensure_browser_healthy here (as this used to, treating ANY nonzero rc
+    # as broken) is wrong on two counts: it misreports a clean defer as a
+    # warning, and ensure_browser_healthy checks the pre-harness
+    # browser-profiles/reddit path (dead since the 2026-05-29 migration), so
+    # it is pure wasted work every time cycles overlap around an active post
+    # (routine — see the "double-fork wrapper" note above BATCH_ID).
+    log "reddit-harness bootstrap deferred (rc=78, poster active elsewhere); skipping this cycle so the poster keeps the tab."
+    python3 "$REPO_DIR/scripts/reddit_browser_lock.py" release 2>/dev/null || true
+    exit 0
+elif [ "$_ensure_rc" != "0" ]; then
+    log "WARNING: reddit-harness bootstrap failed (rc=$_ensure_rc); falling back to ensure_browser_healthy reddit"
     ensure_browser_healthy "reddit"
 fi
 python3 "$REPO_DIR/scripts/reddit_browser_lock.py" release 2>/dev/null || true

package/skill/run-reddit-threads.sh

@@ -256,8 +256,22 @@ export CLAUDE_SESSION_ID=$(uuidgen | tr 'A-Z' 'a-z')
 log "Acquiring reddit-browser lease (TTL 90s, MCP-proxy heartbeated)..."
 python3 "$REPO_DIR/scripts/reddit_browser_lock.py" acquire --timeout 600 --ttl 90 2>&1 | tee -a "$LOG_FILE" || \
     log "WARNING: reddit_browser_lock.py acquire failed; proceeding without lease (peer pipelines may collide)."
-if ! ensure_reddit_browser_for_backend 2>&1 | tee -a "$LOG_FILE"; then
-    log "WARNING: reddit-harness bootstrap failed; falling back to ensure_browser_healthy reddit"
+ensure_reddit_browser_for_backend 2>&1 | tee -a "$LOG_FILE" || true
+_ensure_rc="${PIPESTATUS[0]}"
+if [ "$_ensure_rc" = "78" ]; then
+    # Reserved skip code (harness-common.sh:182-185): the pre-launch hook
+    # deliberately deferred because a poster is mid-drain on the ONE shared
+    # tab. This is NOT a bootstrap failure — falling back to the legacy
+    # ensure_browser_healthy here (as this used to, treating ANY nonzero rc
+    # as broken) is wrong on two counts: it misreports a clean defer as a
+    # warning, and ensure_browser_healthy checks the pre-harness
+    # browser-profiles/reddit path (dead since the 2026-05-29 migration), so
+    # it is pure wasted work every time cycles overlap around an active post
+    # (routine, see the "double-fork wrapper" note in run-reddit-search.sh).
+    log "reddit-harness bootstrap deferred (rc=78, poster active elsewhere); skipping this run so the poster keeps the tab."
+    exit 0
+elif [ "$_ensure_rc" != "0" ]; then
+    log "WARNING: reddit-harness bootstrap failed (rc=$_ensure_rc); falling back to ensure_browser_healthy reddit"
     ensure_browser_healthy "reddit"
 fi
 

package/skill/run-twitter-cycle.sh

@@ -2205,23 +2205,29 @@ SKIP_FILE="/tmp/twitter_cycle_skips_${BATCH_ID}.json"
 if [ -z "${S4L_SANDBOX_CANDIDATES_FILE:-}" ]; then
     python3 "$REPO_DIR/scripts/twitter_batch_phase.py" advance "$BATCH_ID" --phase phase2b-prep 2>&1 | tee -a "$LOG_FILE" || true
 fi
-log "Re-acquiring twitter-browser lock for Phase 2b-prep (read+draft only)..."
-acquire_lock "twitter-browser" 3600 2>>"$LOG_FILE"
-log "twitter-browser lock held (pid=$$) Phase 2b-prep"
-# Drop stale singleton locks (see clean_stale_singleton.sh, also called in Phase 1).
-ensure_twitter_browser_for_backend 2>&1 | tee -a "$LOG_FILE"
-
 # Thread-media capture (2026-06-03, gated by S4L_TWITTER_CAPTURE_MEDIA, default
-# OFF). Now that the browser lock is held and the harness Chrome is up, do ONE
-# cheap deterministic pass over every candidate thread to pull its media
-# (images/videos/GIFs/link-cards), persist each into
+# OFF). When enabled, does ONE cheap deterministic pass over every candidate
+# thread to pull its media (images/videos/GIFs/link-cards), persist each into
 # twitter_candidates.thread_media, and build a MEDIA CONTEXT block injected into
 # the prep prompt so the reply-writer can react to what the tweet visually shows
 # instead of replying text-blind. Must be deterministic (Python pre-fetch) because
 # the prep prompt forbids the model from calling twitter_browser.py. Entirely
 # best-effort: any failure leaves MEDIA_BLOCK empty and the cycle proceeds.
+#
+# The twitter-browser lock exists SOLELY to guard this capture step, so it is
+# now gated by the same flag (2026-07-16): acquiring it unconditionally meant
+# every cycle -- sandbox or live -- contended for the browser lock even when
+# nothing in this phase touches the browser, which is the default state (media
+# capture off). Found live: a sandbox test run waited ~6 minutes in the FIFO
+# lock queue behind a real cycle for a lock it never ended up needing.
 MEDIA_BLOCK=""
 if [ "${S4L_TWITTER_CAPTURE_MEDIA:-0}" = "1" ] || [ "${S4L_TWITTER_CAPTURE_MEDIA:-}" = "true" ]; then
+    log "Re-acquiring twitter-browser lock for Phase 2b-prep (read+draft only)..."
+    acquire_lock "twitter-browser" 3600 2>>"$LOG_FILE"
+    log "twitter-browser lock held (pid=$$) Phase 2b-prep"
+    # Drop stale singleton locks (see clean_stale_singleton.sh, also called in Phase 1).
+    ensure_twitter_browser_for_backend 2>&1 | tee -a "$LOG_FILE"
+
     if [ -s "$MEDIA_URLS_FILE" ]; then
         log "Phase 2b-prep: capturing thread media for $(wc -l < "$MEDIA_URLS_FILE" | tr -d ' ') candidate(s)..."
         MEDIA_BLOCK=$(python3 "$REPO_DIR/scripts/capture_thread_media.py" --urls-file "$MEDIA_URLS_FILE" --scroll 1 2>>"$LOG_FILE" || true)
@@ -2231,28 +2237,29 @@ if [ "${S4L_TWITTER_CAPTURE_MEDIA:-0}" = "1" ] || [ "${S4L_TWITTER_CAPTURE_MEDIA
             log "Phase 2b-prep: no media captured (none found or capture skipped)."
         fi
     fi
+    rm -f "$MEDIA_URLS_FILE" 2>/dev/null || true
+
+    # Release the twitter-browser lock now. Thread-media capture above was the
+    # ONLY browser-touching step in Phase 2b-prep; the Claude drafting call below
+    # is architecturally browser-free (--strict-mcp-config omits the
+    # twitter-harness MCP, so the model can never reach the CDP Chrome even if it
+    # tried) and, since the 2026-06-23 queue migration, "run-twitter-cycle-prep"
+    # routes through claude_job.py to an independent worker process that can
+    # block for up to S4L_CLAUDE_QUEUE_TIMEOUT (1800s default). Holding the
+    # browser lock across that wait made this process a preemption target for
+    # any post that needed the browser: the post-vs-scan hijack fix SIGKILLs
+    # whoever holds the lock (see docs/twitter_browser_lock.md), which killed
+    # this process mid-wait while the worker kept drafting in the background,
+    # stranding the finished result until the salvage reconciler's 35-min window
+    # (scripts/salvage_orphaned_prep_results.py) picked it up. Releasing here
+    # removes this whole phase from being a SIGKILL target for the rest of the
+    # drafting wait. Phase 2b-post re-acquires unconditionally below.
+    log "Releasing twitter-browser lock before Claude drafting (drafting never touches the browser)..."
+    release_lock "twitter-browser" 2>>"$LOG_FILE"
 else
-    log "Phase 2b-prep: thread-media capture disabled (S4L_TWITTER_CAPTURE_MEDIA not set)."
+    log "Phase 2b-prep: thread-media capture disabled (S4L_TWITTER_CAPTURE_MEDIA not set); skipping the twitter-browser lock entirely."
+    rm -f "$MEDIA_URLS_FILE" 2>/dev/null || true
 fi
-rm -f "$MEDIA_URLS_FILE" 2>/dev/null || true
-
-# Release the twitter-browser lock now. Thread-media capture above was the
-# ONLY browser-touching step in Phase 2b-prep; the Claude drafting call below
-# is architecturally browser-free (--strict-mcp-config omits the
-# twitter-harness MCP, so the model can never reach the CDP Chrome even if it
-# tried) and, since the 2026-06-23 queue migration, "run-twitter-cycle-prep"
-# routes through claude_job.py to an independent worker process that can
-# block for up to S4L_CLAUDE_QUEUE_TIMEOUT (1800s default). Holding the
-# browser lock across that wait made this process a preemption target for
-# any post that needed the browser: the post-vs-scan hijack fix SIGKILLs
-# whoever holds the lock (see docs/twitter_browser_lock.md), which killed
-# this process mid-wait while the worker kept drafting in the background,
-# stranding the finished result until the salvage reconciler's 35-min window
-# (scripts/salvage_orphaned_prep_results.py) picked it up. Releasing here
-# removes this whole phase from being a SIGKILL target for the rest of the
-# drafting wait. Phase 2b-post re-acquires unconditionally below.
-log "Releasing twitter-browser lock before Claude drafting (drafting never touches the browser)..."
-release_lock "twitter-browser" 2>>"$LOG_FILE"
 
 # --- ACCOUNT VOICE CORPUS injection (both lanes, 2026-07-15) -----------------
 # build_persona.py apply / voice_exemplars.py write a raw first-hand corpus
@@ -2824,6 +2831,17 @@ if [ "${DRAFT_ONLY:-0}" = "1" ]; then
         --tweets-pulled "${TWEETS_PULLED:-0}" --candidates "${BATCH_COUNT:-0}" --above-floor "${HIGH_DELTA_COUNT:-0}" \
         --cost "$_COST" --elapsed $(( $(date +%s) - RUN_START )) 2>/dev/null || true
     _SA_RUN_SUMMARY_EMITTED=1
+    # Sandbox runs have no MCP wrapper watching stdout for this marker to hand
+    # the plan off to merge_review_queue.py (that handoff normally happens
+    # outside this script), so a sandbox plan would otherwise just rot on
+    # disk -- found live 2026-07-15, had to hand-reconstruct one after the
+    # fact. $PLAN_FILE already carries the fully-transformed candidates
+    # (reply_text/drafts/experiments.sandbox stamp, same shape any real draft
+    # gets), so this is a straight, no-transform call into the same merge
+    # step everything else uses.
+    if [ -n "${S4L_SANDBOX_CANDIDATES_FILE:-}" ]; then
+        python3 "$REPO_DIR/scripts/merge_review_queue.py" --plan "$PLAN_FILE" 2>&1 | tee -a "$LOG_FILE"
+    fi
     # IMPORTANT: do NOT rm -f "$PLAN_FILE" here; the reviewer needs it. Print a
     # machine-readable marker so the MCP wrapper can locate the plan deterministically.
     echo "DRAFT_ONLY_PLAN=$PLAN_FILE"