social-autoposter 1.6.150 → 1.6.152

package/mcp/dist/index.js

@@ -2077,7 +2077,7 @@ async function autopilotLoaded() {
 // fires every minute, claims ONE job, runs the pipeline's own prompt as its
 // Claude turn, writes the result back, and stops.
 // ===========================================================================
-const QUEUE_WORKER_PROMPT_VERSION = 5; // v5: drop the per-run draft cap (v3/v4). The ~90s kills were NOT a drafting-time limit; they were unattended permission-prompt stalls on un-auto-approved Bash calls. Real fix = blanket Bash allow in queueWorkerAllowedTools so no prompt is ever emitted. Worker drafts the FULL set again.
+const QUEUE_WORKER_PROMPT_VERSION = 6; // v6: the ~90s kill is a host INACTIVITY watchdog (resets on each tool call), not a fixed cap. The worker was drafting all candidates in ONE silent turn (its HARD RULES even forbade log_draft.py), starving the clock -> killed mid-draft. v6 requires incremental draft-then-log per candidate and allows the log_draft.py persist commands, so per-candidate tool calls keep the session alive and the full set drains.
 const QUEUE_WORKER_PROMPT_MARKER = "saps_queue_worker_prompt_version";
 // One spec per worker task. queueType MUST match scripts/claude_job.py TAG_TO_TYPE.
 const QUEUE_WORKERS = [
@@ -2180,6 +2180,13 @@ function queueWorkerBody(spec) {
     const job = path.join(repoDir(), "scripts", "claude_job.py");
     const sd = sapsStateDir();
     const outDir = queueDir();
+    // The Phase-2b "twitter-prep" worker drafts replies for several candidates. The
+    // host kills an unattended scheduled session ~90s after its LAST tool call (an
+    // inactivity watchdog). Drafting everything in one silent turn starves that clock
+    // and the session is killed mid-draft having submitted nothing. So it MUST work
+    // incrementally — draft + run log_draft.py per candidate — which persists drafts
+    // (the pipeline already wants this) AND keeps the activity clock alive.
+    const isDraft = spec.queueType === "twitter-prep";
     return [
         `You are the S4L "${spec.human}" queue worker. Run ONE iteration, then STOP.`,
         ``,
@@ -2189,6 +2196,23 @@ function queueWorkerBody(spec) {
             `You do this with Bash and Write, and NOTHING else. This run is unattended — ` +
             `reaching for any other tool, or trying to "investigate", STALLS it forever.`,
         ``,
+        ...(isDraft
+            ? [
+                `PACING — CRITICAL: this unattended session is terminated ~90 seconds after ` +
+                    `your LAST tool call (a host inactivity timeout). The prompt below asks ` +
+                    `you to draft replies for SEVERAL candidates. Do NOT draft them all ` +
+                    `silently in one turn — that emits no tool calls, the clock expires, and ` +
+                    `you are KILLED mid-draft having submitted NOTHING. Work ONE candidate at ` +
+                    `a time: draft its reply, then IMMEDIATELY run that candidate's ` +
+                    `log_draft.py command exactly as the prompt's persist step (3a) specifies ` +
+                    `(a quick Bash call), THEN move to the next. Those per-candidate Bash ` +
+                    `calls are what keep the session alive. Begin the first candidate ` +
+                    `promptly — do NOT spend a long silent stretch analysing before your ` +
+                    `first tool call. Only after EVERY candidate is drafted and logged do you ` +
+                    `do step 3.`,
+                ``,
+            ]
+            : []),
         `Steps:`,
         `1. Claim the next job. Run this EXACT Bash command:`,
         `     ${py} ${job} next --type ${spec.queueType} --prompt-file --state-dir ${sd}`,
@@ -2211,11 +2235,18 @@ function queueWorkerBody(spec) {
         `4. Report in ONE short line what you did, then STOP. Do NOT claim another job, ` +
             `do NOT loop, do NOT read other files, do NOT call any other tool.`,
         ``,
-        `HARD RULES: ONLY the Bash tool (to run claude_job.py), the Read tool (to read ` +
-            `the prompt/schema sidecar files), and the Write tool (to write the result ` +
-            `file). NEVER run any other shell command. NEVER edit, post, or touch anything ` +
-            `else. An empty queue is the NORMAL, expected case most minutes — it is success, ` +
-            `not a problem to debug.`,
+        isDraft
+            ? `HARD RULES: use ONLY the Bash tool (to run claude_job.py AND the per-candidate ` +
+                `log_draft.py persist commands the prompt tells you to run), the Read tool (the ` +
+                `prompt/schema sidecar + the SKILL/config files the prompt names), and the Write ` +
+                `tool (the result file). NEVER post, reply, edit a draft, open a browser, or run ` +
+                `any command the prompt does not explicitly give you. An empty queue is the ` +
+                `NORMAL, expected case most minutes — it is success, not a problem to debug.`
+            : `HARD RULES: ONLY the Bash tool (to run claude_job.py), the Read tool (to read ` +
+                `the prompt/schema sidecar files), and the Write tool (to write the result ` +
+                `file). NEVER run any other shell command. NEVER edit, post, or touch anything ` +
+                `else. An empty queue is the NORMAL, expected case most minutes — it is success, ` +
+                `not a problem to debug.`,
     ].join("\n");
 }
 // Full canonical SKILL.md (frontmatter + body + version marker) the MCP writes

package/mcp/dist/version.json

@@ -1,4 +1,4 @@
 {
-  "version": "1.6.150",
-  "installedAt": "2026-06-29T22:17:40.330Z"
+  "version": "1.6.152",
+  "installedAt": "2026-06-29T22:42:31.815Z"
 }

package/mcp/manifest.json

@@ -2,7 +2,7 @@
   "dxt_version": "0.1",
   "name": "social-autoposter",
   "display_name": "S4L",
-  "version": "1.6.150",
+  "version": "1.6.152",
   "description": "Draft, review, approve, and autopilot X/Twitter posts.",
   "long_description": "The disclaimer above is generic Claude boilerplate. S4L 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 end to end**\n\n2\\. Quit fully with CMD+Q, restart Claude, and paste the prompt into a new chat.",
   "author": {

package/mcp/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@m13v/social-autoposter-mcp",
-  "version": "1.6.150",
+  "version": "1.6.152",
   "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": "social-autoposter",
-  "version": "1.6.150",
+  "version": "1.6.152",
   "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/skill/run-twitter-cycle.sh

@@ -1772,21 +1772,23 @@ log "Engagement style assigned: mode=$PICKED_MODE style=${PICKED_STYLE:-(invent)
 # The arm is stamped onto every post this cycle via SAPS_DRAFT_PROMPT_VARIANT
 # (read by twitter_post_plan.py -> log_post.py -> posts.draft_prompt_variant),
 # mirroring the tail_link_variant plumbing. Split tunable via
-# TWITTER_DRAFT_PROMPT_AB_RATE (fraction assigned to 'treatment'). DEFAULT 0 =
-# 100% control = experiment OFF (no behavior change anywhere until explicitly
-# enabled, mirroring how the "written with ai" tag is off by default). Set it to
-# e.g. 0.5 in .env to activate a 50/50 split. The dashboard reads the SAME var
+# TWITTER_DRAFT_PROMPT_AB_RATE = fraction of cycles assigned to 'treatment' (the
+# decoupled directive). CODE DEFAULT 1 = 100% treatment: a fresh plugin install /
+# new user with no override drafts in the DECOUPLED style by default (the old
+# concede->pivot 'control' is opt-in, not the default for customers). OUR own
+# install pins this to 0.5 in .env for a 50/50 holdback so we can measure
+# decoupled vs the old behavior on our account. The dashboard reads the SAME var
 # with the SAME default (bin/server.js), so display and routing never diverge.
-DRAFT_PROMPT_AB_RATE="${TWITTER_DRAFT_PROMPT_AB_RATE:-0}"
+DRAFT_PROMPT_AB_RATE="${TWITTER_DRAFT_PROMPT_AB_RATE:-1}"
 SAPS_DRAFT_PROMPT_VARIANT=$(python3 -c "
 import random
 try:
     rate = float('$DRAFT_PROMPT_AB_RATE')
 except Exception:
-    rate = 0.0
+    rate = 1.0
 rate = min(1.0, max(0.0, rate))
 print('treatment' if random.random() < rate else 'control')
-" 2>/dev/null || echo control)
+" 2>/dev/null || echo treatment)
 export SAPS_DRAFT_PROMPT_VARIANT
 log "Draft-prompt A/B arm: $SAPS_DRAFT_PROMPT_VARIANT (rate=$DRAFT_PROMPT_AB_RATE)"
 if [ "$SAPS_DRAFT_PROMPT_VARIANT" = "treatment" ]; then

package/scripts/_scan_pending_detail.py

@@ -1,33 +0,0 @@
-import json, sys
-sys.path.insert(0, "scripts")
-import http_api
-
-resp = http_api.api_get("/api/v1/dms/pending")
-rows = (resp.get("data") or {}).get("pending") or []
-ids = [r["id"] for r in rows]
-out = []
-for dm_id in ids:
-    r = http_api.api_get(f"/api/v1/dms/{dm_id}", ok_on_404=True)
-    dm = (r.get("data") or {}).get("dm") or {}
-    mr = http_api.api_get(f"/api/v1/dms/{dm_id}/messages")
-    msgs = (mr.get("data") or {}).get("messages") or []
-    hist = [{"d": m.get("direction"), "a": m.get("author"), "c": (m.get("content") or "")} for m in msgs]
-    out.append({
-        "id": dm_id,
-        "platform": dm.get("platform"),
-        "author": dm.get("their_author"),
-        "surface": dm.get("surface"),
-        "chat_url": dm.get("chat_url"),
-        "tier": dm.get("tier"),
-        "qual": dm.get("qualification_status"),
-        "mode": dm.get("mode"),
-        "target_project": dm.get("target_project"),
-        "project_name": dm.get("project_name"),
-        "interest": dm.get("interest_level"),
-        "origin_public_comment_url": dm.get("origin_public_comment_url"),
-        "our_prior_public_reply_url": dm.get("our_prior_public_reply_url"),
-        "booking_sent": dm.get("booking_link_sent_at"),
-        "msgcount": dm.get("message_count"),
-        "hist": hist,
-    })
-print(json.dumps(out, indent=1, ensure_ascii=False))