@mmerterden/multi-agent-pipeline 12.4.0 → 12.6.0

package/pipeline/scripts/smoke-command-inventory.sh

@@ -0,0 +1,81 @@
+#!/usr/bin/env bash
+# smoke-command-inventory.sh  -  the command inventory is DERIVED, never asserted.
+#
+# Every count that describes "how many multi-agent commands exist" used to be a
+# literal in prose (contract header, sync skill, dispatcher). Adding a command
+# left those literals stale while every gate stayed green, because the gates
+# asserted the literal instead of the tree. This gate computes the count from
+# pipeline/commands/multi-agent/*/ and requires the prose to agree.
+#
+# Verifies:
+#   1. contract header count == number of command directories
+#   2. every command directory appears in the contract inventory block
+#   3. both sync surfaces (command + Copilot skill) carry the same count twice
+#      ("N commands are synced" + "instructions + N sub-command skills")
+#   4. every command directory appears in both sync inventory blocks
+#   5. every command directory has its shared/core skill counterpart
+#
+# Exit: 0 all green, 1 any failure.
+
+set -euo pipefail
+
+HERE="$(cd "$(dirname "$0")" && pwd)"
+ROOT="$(cd "$HERE/../.." && pwd)"
+CMD_DIR="$ROOT/pipeline/commands/multi-agent"
+CONTRACT="$ROOT/pipeline/multi-agent-refs/cross-cli-contract.md"
+SYNC="$ROOT/pipeline/commands/multi-agent/sync/SKILL.md"
+SYNC_SKILL="$ROOT/pipeline/skills/shared/core/multi-agent-sync/SKILL.md"
+SKILLS_DIR="$ROOT/pipeline/skills/shared/core"
+
+PASS=0
+FAIL=0
+pass() { PASS=$((PASS+1)); echo "  ✓ $1"; }
+fail() { FAIL=$((FAIL+1)); echo "  ✗ $1"; }
+
+echo "══ smoke-command-inventory ══"
+
+for f in "$CONTRACT" "$SYNC" "$SYNC_SKILL"; do
+  [ -f "$f" ] || { echo "  ✗ missing file: ${f#$ROOT/}"; exit 1; }
+done
+
+COMMANDS=$(find "$CMD_DIR" -mindepth 1 -maxdepth 1 -type d -exec basename {} \; | sort)
+N=$(printf '%s\n' "$COMMANDS" | grep -c . || true)
+
+echo "→ 1. derived count from the command tree: $N commands"
+[ "$N" -gt 0 ] && pass "found $N command directories" || fail "no command directories under ${CMD_DIR#$ROOT/}"
+
+echo "→ 2. contract header agrees"
+if grep -Fq "## 1. Command Inventory ($N commands)" "$CONTRACT"; then
+  pass "contract header = $N"
+else
+  fail "contract header does not say ($N commands): $(grep -o '## 1\. Command Inventory ([0-9]* commands)' "$CONTRACT" || echo 'header not found')"
+fi
+
+echo "→ 3. sync surfaces agree (2 counts x 2 files)"
+for f in "$SYNC" "$SYNC_SKILL"; do
+  rel="${f#$ROOT/}"
+  grep -Fq "**$N commands are synced**" "$f" && pass "$rel: synced count = $N" || fail "$rel: 'N commands are synced' != $N"
+  grep -Fq "instructions + $N sub-command skills" "$f" && pass "$rel: step-2 count = $N" || fail "$rel: 'instructions + N sub-command skills' != $N"
+done
+
+echo "→ 4. every command is listed in the contract + both sync inventories"
+MISSING_LIST=0
+while IFS= read -r cmd; do
+  [ -n "$cmd" ] || continue
+  for f in "$CONTRACT" "$SYNC" "$SYNC_SKILL"; do
+    grep -Eq "(^|[ ,])$cmd([ ,]|$)" "$f" || { fail "${f#$ROOT/}: inventory missing '$cmd'"; MISSING_LIST=1; }
+  done
+done <<< "$COMMANDS"
+[ "$MISSING_LIST" -eq 0 ] && pass "all $N commands listed on all 3 surfaces"
+
+echo "→ 5. shared/core skill counterpart exists for every command"
+MISSING_SKILL=0
+while IFS= read -r cmd; do
+  [ -n "$cmd" ] || continue
+  [ -f "$SKILLS_DIR/multi-agent-$cmd/SKILL.md" ] || { fail "no shared/core skill for '$cmd'"; MISSING_SKILL=1; }
+done <<< "$COMMANDS"
+[ "$MISSING_SKILL" -eq 0 ] && pass "all $N commands have a shared/core skill"
+
+echo ""
+echo "══ command-inventory smoke: $PASS passed, $FAIL failed ══"
+[ "$FAIL" -eq 0 ] || exit 1

package/pipeline/scripts/smoke-commands-skills-parity.sh

@@ -36,7 +36,7 @@ fail() { FAIL=$((FAIL+1)); echo "  ✗ $1"; }
 # Underscore-prefixed files (e.g. _account-picker.md) are internal fragments
 # loaded by other commands via Read; they're not invocable slash commands and
 # intentionally have no matching skill.
-CMDS=$(find "$CMDS_DIR" -mindepth 2 -maxdepth 2 -name "SKILL.md" -exec dirname {} \; | xargs -n1 basename | sort)
+CMDS=$(find "$CMDS_DIR" -mindepth 2 -maxdepth 2 -name "SKILL.md" -exec sh -c 'for f do basename "$(dirname "$f")"; done' sh {} + | sort)
 
 # help.md in commands maps to multi-agent-help skill; the main "multi-agent"
 # skill covers the behavior documented across multiple command files, so we

package/pipeline/scripts/smoke-compliance-skills.sh

@@ -5,7 +5,7 @@
 # that the 4 claimed consumer surfaces actually reference them (not just promise).
 #
 # Covers:
-#   1. apple-archive-compliance/SKILL.md exists, has frontmatter, 17-rule catalog,
+#   1. apple-archive-compliance/SKILL.md exists, has frontmatter, 18-rule catalog,
 #      EN + TR humanizer blocks, severity table, prerequisite detection.
 #   2. google-play-compliance/SKILL.md exists, has frontmatter, 21-rule catalog
 #      across 4 categories, EN + TR humanizer blocks, severity table, prereq detection.
@@ -43,8 +43,8 @@ grep -q "^name: apple-archive-compliance$" "$APPLE_SKILL" && pass "apple frontma
 grep -q "^user-invocable: true$" "$APPLE_SKILL" && pass "apple user-invocable" || fail "apple user-invocable"
 grep -q "argument-hint:" "$APPLE_SKILL" && pass "apple argument-hint" || fail "apple argument-hint"
 # Count rule rows in the catalog table (excluding header + separator)
-APPLE_RULES=$(awk '/^## 17-Rule Catalog/,/^## Severity mapping/' "$APPLE_SKILL" | grep -cE '^\| *[0-9]+ *\|' || true)
-[ "$APPLE_RULES" -eq 17 ] && pass "apple catalog has 17 rule rows" || fail "apple catalog has $APPLE_RULES rule rows (want 17)"
+APPLE_RULES=$(awk '/^## 18-Rule Catalog/,/^## Severity mapping/' "$APPLE_SKILL" | grep -cE '^\| *[0-9]+ *\|' || true)
+[ "$APPLE_RULES" -eq 18 ] && pass "apple catalog has 18 rule rows" || fail "apple catalog has $APPLE_RULES rule rows (want 18)"
 
 echo "→ 2. apple skill humanizer blocks EN + TR + severity table"
 # v8.2 split humanizer language from picker language: gating moved from
@@ -59,7 +59,7 @@ echo "→ 3. apple skill prerequisite detection + archive resolution (v8.4.0+ us
 grep -q 'ios_app_store_audit' "$APPLE_SKILL" && pass "apple references ios_app_store_audit MCP tool" || fail "apple ios_app_store_audit reference missing"
 grep -q '.xcarchive' "$APPLE_SKILL" && pass "apple references .xcarchive" || fail "apple .xcarchive missing"
 grep -q '@mmerterden/dev-toolkit-mcp' "$APPLE_SKILL" && pass "apple references @mmerterden/dev-toolkit-mcp package" || fail "apple dev-toolkit-mcp reference missing"
-grep -q 'mcp__dev_toolkit__ios_app_store_audit' "$APPLE_SKILL" && pass "apple shows MCP-call invocation form" || fail "apple MCP-call invocation form missing"
+grep -q 'mcp__dev-toolkit__ios_app_store_audit' "$APPLE_SKILL" && pass "apple shows MCP-call invocation form" || fail "apple MCP-call invocation form missing"
 
 echo "→ 4. google-play-compliance SKILL.md frontmatter + catalog shape"
 grep -q "^name: google-play-compliance$" "$GOOGLE_SKILL" && pass "google frontmatter name" || fail "google frontmatter name"

package/pipeline/scripts/smoke-cross-cli-behavior.sh

@@ -61,6 +61,16 @@ fi
 # ──────────────────────────────────────────────────────────────────────────
 echo ""
 echo "→ 2. No deprecated placeholders in Claude source"
+# Corpus guard: the scans below can only be trusted if there is something to
+# scan. Without this, an empty or missing command tree reports every pattern
+# clean instead of reporting that it found nothing to check.
+SKILL_COUNT=$(find "$CLAUDE_CMDS" -name SKILL.md 2>/dev/null | wc -l | tr -d ' ')
+if [ "$SKILL_COUNT" -gt 0 ]; then
+  pass "command corpus non-empty ($SKILL_COUNT SKILL.md)"
+else
+  fail "no SKILL.md under $CLAUDE_CMDS  -  placeholder and hardcode scans below are vacuous"
+fi
+
 # Per refs/cross-cli-contract.md Section 2.5
 DEPRECATED=(
   "{github-username}"
@@ -72,7 +82,7 @@ DEPRECATED=(
 )
 for p in "${DEPRECATED[@]}"; do
   # grep -l returns filenames; empty = no matches
-  FOUND=$(grep -l -F -- "$p" $(find "$CLAUDE_CMDS" -name SKILL.md 2>/dev/null) 2>/dev/null || true)
+  FOUND=$(find "$CLAUDE_CMDS" -name SKILL.md -exec grep -l -F -- "$p" {} + 2>/dev/null || true)
   if [ -z "$FOUND" ]; then
     pass "no '$p' in Claude commands"
   else
@@ -88,7 +98,7 @@ PERSONAL_HARDCODES=(
   "${USER}_tkgithub"
 )
 for p in "${PERSONAL_HARDCODES[@]}"; do
-  FOUND=$(grep -l -F -- "$p" $(find "$CLAUDE_CMDS" -name SKILL.md 2>/dev/null) 2>/dev/null || true)
+  FOUND=$(find "$CLAUDE_CMDS" -name SKILL.md -exec grep -l -F -- "$p" {} + 2>/dev/null || true)
   if [ -z "$FOUND" ]; then
     pass "no '$p' leaked into Claude commands"
   else

package/pipeline/scripts/smoke-generate-issue.sh

@@ -9,7 +9,7 @@
 #   5. Convention mining + createmeta + Agile sprint endpoints present.
 #   6. UTF-8 verbatim POST path (--rawfile + --data-binary) present.
 #   7. Standard template vocab: Task/Story (Scope/Acceptance/Test Scenarios), Bug (Steps/Expected/Actual/Environment), Swagger.
-#   8. Registration: dispatcher rows, inventory lists, count consistency (36).
+#   8. Registration: dispatcher rows, inventory lists, count derived from the command tree.
 #   9. Forbidden strings absent (em-dash, section sign, the M-word, unicode_escape).
 
 set -euo pipefail
@@ -21,6 +21,7 @@ REF="$ROOT/pipeline/multi-agent-refs/generate-issue.md"
 SKILL="$ROOT/pipeline/skills/shared/core/multi-agent-create-jira/SKILL.md"
 DISPATCHER="$ROOT/pipeline/commands/multi-agent/SKILL.md"
 CONTRACT="$ROOT/pipeline/multi-agent-refs/cross-cli-contract.md"
+CMD_COUNT=$(find "$ROOT/pipeline/commands/multi-agent" -mindepth 1 -maxdepth 1 -type d | wc -l | tr -d ' ')
 SYNC="$ROOT/pipeline/commands/multi-agent/sync/SKILL.md"
 HELP="$ROOT/pipeline/commands/multi-agent/help/SKILL.md"
 
@@ -88,14 +89,14 @@ grep -Fqi "auto-sizing" "$REF" && pass "auto-sizing rule present" || fail "auto-
 grep -Fqi "swagger" "$REF" && pass "swagger context present" || fail "swagger context missing"
 
 echo ""
-echo "→ 8. registration + count consistency (36)"
+echo "→ 8. registration + count consistency (derived)"
 grep -Fq '| `create-jira ' "$DISPATCHER" && pass "dispatcher routing: create-jira" || fail "dispatcher routing missing: generate"
 grep -Fq "create-jira/SKILL.md" "$DISPATCHER" && pass "dispatcher loading row" || fail "dispatcher loading row missing"
-grep -Fq "## 1. Command Inventory (41 commands)" "$CONTRACT" && pass "contract count = 41" || fail "contract count wrong"
+grep -Fq "## 1. Command Inventory ($CMD_COUNT commands)" "$CONTRACT" && pass "contract count = $CMD_COUNT (derived)" || fail "contract count wrong"
 grep -Eq '(^|[ ,])create-jira([ ,]|$)' "$CONTRACT" && pass "contract list has create-jira" || fail "contract list missing generate"
-grep -Fq "**41 commands are synced**" "$SYNC" && pass "sync count = 41" || fail "sync count wrong"
+grep -Fq "**$CMD_COUNT commands are synced**" "$SYNC" && pass "sync count = $CMD_COUNT (derived)" || fail "sync count wrong"
 SYNC_SKILL="$ROOT/pipeline/skills/shared/core/multi-agent-sync/SKILL.md"
-grep -Fq "**41 commands are synced**" "$SYNC_SKILL" && pass "sync SKILL count = 41" || fail "sync SKILL count wrong (Copilot parity)"
+grep -Fq "**$CMD_COUNT commands are synced**" "$SYNC_SKILL" && pass "sync SKILL count = $CMD_COUNT (derived)" || fail "sync SKILL count wrong (Copilot parity)"
 grep -Fq "create-jira" "$ROOT/pipeline/skills/shared/core/multi-agent-help/SKILL.md" && pass "help SKILL lists create-jira" || fail "help SKILL missing create-jira (Copilot parity)"
 grep -c "multi-agent:create-jira" "$HELP" | awk '{exit !($1>=2)}' && pass "help.md has EN + TR entries" || fail "help.md entries incomplete"
 # no stale split-command references anywhere in the synced surfaces

package/pipeline/scripts/smoke-learning-curve.sh

@@ -0,0 +1,61 @@
+#!/usr/bin/env bash
+# smoke-learning-curve.sh  -  contract for learning-curve.mjs (time-bucketed
+# trend over metrics.jsonl).
+
+set -uo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+LC="$SCRIPT_DIR/learning-curve.mjs"
+
+PASS=0
+FAIL=0
+record_pass() { PASS=$((PASS + 1)); printf '  \342\234\223 %s\n' "$1"; }
+record_fail() { FAIL=$((FAIL + 1)); printf '  \342\234\227 %s\n' "$1"; }
+
+printf '\342\206\222 smoke-learning-curve: metrics trend contract\n'
+
+tmp=$(mktemp -d "${TMPDIR:-/tmp}/lc.XXXXXX")
+trap 'rm -rf "$tmp"' EXIT
+M="$tmp/metrics.jsonl"
+
+# Two early tasks (noisy: 3/2 review cycles, low cache) + two later tasks
+# (clean: 1 cycle, high cache) -> the curve should improve.
+cat > "$M" <<'EOF'
+{"ts":"2026-06-02T10:00:00Z","task_id":"t1","phase":"4","event":"review.completed","details":{"review_cycles":3,"tokens_in":100000,"tokens_out":8000,"tokens_cached":10000}}
+{"ts":"2026-06-02T10:05:00Z","task_id":"t1","phase":"3","event":"rework.started","details":{}}
+{"ts":"2026-06-03T09:00:00Z","task_id":"t2","phase":"4","event":"review.completed","details":{"review_cycles":2,"tokens_in":90000,"tokens_out":7000,"tokens_cached":20000}}
+{"ts":"2026-06-24T09:00:00Z","task_id":"t3","phase":"4","event":"review.completed","details":{"review_cycles":1,"tokens_in":40000,"tokens_out":4000,"tokens_cached":30000}}
+{"ts":"2026-06-25T09:00:00Z","task_id":"t4","phase":"4","event":"review.completed","details":{"review_cycles":1,"tokens_in":38000,"tokens_out":3500,"tokens_cached":32000}}
+EOF
+
+# 1. parses
+bash -n "$SCRIPT_DIR/smoke-learning-curve.sh" >/dev/null 2>&1 || true
+if node --check "$LC" 2>/dev/null; then record_pass "learning-curve.mjs parses"; else record_fail "syntax error"; fi
+
+# 2. JSON: buckets + improving trend
+out=$(METRICS_FILE="$M" node "$LC" --json 2>/dev/null)
+nbuckets=$(printf '%s' "$out" | python3 -c "import json,sys;print(len(json.load(sys.stdin)['buckets']))")
+[ "$nbuckets" -ge 2 ] && record_pass "produces >=2 time buckets" || record_fail "expected >=2 buckets, got $nbuckets"
+
+tokens_delta=$(printf '%s' "$out" | python3 -c "import json,sys;print(json.load(sys.stdin)['trend']['tokensPerTask'])")
+case "$tokens_delta" in
+  -*) record_pass "tokens/task trend is negative (improving)" ;;
+  *) record_fail "tokens/task trend should drop, got $tokens_delta" ;;
+esac
+cache_delta=$(printf '%s' "$out" | python3 -c "import json,sys;print(json.load(sys.stdin)['trend']['cacheRatio'])")
+case "$cache_delta" in
+  -*|0|0.0) record_fail "cache ratio should rise, got $cache_delta" ;;
+  *) record_pass "cache ratio trend rises (improving)" ;;
+esac
+
+# 3. markdown renders a table + trend
+md=$(METRICS_FILE="$M" node "$LC" 2>/dev/null)
+printf '%s' "$md" | grep -q "Learning curve" && record_pass "markdown has a header" || record_fail "no markdown header"
+printf '%s' "$md" | grep -q "Trend" && record_pass "markdown shows a trend block" || record_fail "no trend block"
+
+# 4. missing metrics -> graceful, exit 0
+METRICS_FILE="/nope/none.jsonl" node "$LC" --json >/dev/null 2>&1
+[ "$?" -eq 0 ] && record_pass "missing metrics is a graceful no-op (exit 0)" || record_fail "missing metrics should exit 0"
+
+printf '\n\342\225\220\342\225\220 smoke-learning-curve: %d passed, %d failed \342\225\220\342\225\220\n' "$PASS" "$FAIL"
+[ "$FAIL" -eq 0 ]

package/pipeline/scripts/smoke-per-repo-memory.sh

@@ -35,7 +35,7 @@ repo="$sandbox/repo"
 mkdir -p "$repo"
 
 # --- 1: pref off → silent exit 0 ---
-printf '{"global":{"perRepoMemory":false}}' > "$HOME/.claude/preferences.json"
+printf '{"global":{"perRepoMemory":false}}' > "$HOME/.claude/multi-agent-preferences.json"
 out=$(bash "$LOAD" "$repo" 2>&1 || true)
 if [ -z "$out" ]; then
   record_pass "load exits silent when pref off"
@@ -44,7 +44,7 @@ else
 fi
 
 # --- 2: pref on + no memory → silent exit 0 ---
-printf '{"global":{"perRepoMemory":true}}' > "$HOME/.claude/preferences.json"
+printf '{"global":{"perRepoMemory":true}}' > "$HOME/.claude/multi-agent-preferences.json"
 out=$(bash "$LOAD" "$repo" 2>&1 || true)
 if [ -z "$out" ]; then
   record_pass "load exits silent when pref on but no memory present"

package/pipeline/scripts/smoke-review-readiness.sh

@@ -21,6 +21,7 @@ SK_J="$ROOT/pipeline/skills/shared/core/multi-agent-review-jira/SKILL.md"
 SK_I="$ROOT/pipeline/skills/shared/core/multi-agent-review-issue/SKILL.md"
 DISPATCHER="$ROOT/pipeline/commands/multi-agent/SKILL.md"
 CONTRACT="$ROOT/pipeline/multi-agent-refs/cross-cli-contract.md"
+CMD_COUNT=$(find "$ROOT/pipeline/commands/multi-agent" -mindepth 1 -maxdepth 1 -type d | wc -l | tr -d ' ')
 HELP="$ROOT/pipeline/commands/multi-agent/help/SKILL.md"
 
 PASS=0; FAIL=0
@@ -61,12 +62,12 @@ grep -Fqi "confirm" "$REF" && pass "ref has confirm-before-post gate" || fail "r
 grep -Fqi "rubric" "$REF" && pass "ref has readiness rubric" || fail "ref missing rubric"
 
 echo ""
-echo "→ 5. registration + count (38)"
+echo "→ 5. registration + count (derived)"
 grep -Fq '| `review-jira ' "$DISPATCHER" && pass "dispatcher desc: review-jira" || fail "dispatcher desc missing review-jira"
 grep -Fq '| `review-issue ' "$DISPATCHER" && pass "dispatcher desc: review-issue" || fail "dispatcher desc missing review-issue"
 grep -Fq "review-jira/SKILL.md" "$DISPATCHER" && pass "dispatcher routing: review-jira" || fail "dispatcher routing missing review-jira"
 grep -Fq "review-issue/SKILL.md" "$DISPATCHER" && pass "dispatcher routing: review-issue" || fail "dispatcher routing missing review-issue"
-grep -Fq "## 1. Command Inventory (41 commands)" "$CONTRACT" && pass "contract count = 41" || fail "contract count wrong"
+grep -Fq "## 1. Command Inventory ($CMD_COUNT commands)" "$CONTRACT" && pass "contract count = $CMD_COUNT (derived)" || fail "contract count wrong"
 grep -Eq '(^|[ ,])review-jira([ ,]|$)' "$CONTRACT" && pass "contract inventory has review-jira" || fail "contract missing review-jira"
 grep -Eq '(^|[ ,])review-issue([ ,]|$)' "$CONTRACT" && pass "contract inventory has review-issue" || fail "contract missing review-issue"
 grep -Fq "multi-agent:review-jira" "$HELP" && pass "help lists review-jira" || fail "help missing review-jira"

package/pipeline/scripts/smoke-schema-validation.sh

@@ -87,17 +87,31 @@ done
 
 # ──────────────────────────────────────────────────────────────────────────
 echo ""
-echo "→ 4. prefs.schema.json  -  keychainMapping has 12 services"
+echo "→ 4. prefs.schema.json  -  keychainMapping covers every resolved service"
+# A floor, not a pin: new credentials get added over time, and pinning the count
+# made every addition a false failure. What matters is that the keys shipped
+# flows resolve are all declared.
 KM_COUNT=$(node -e "
   const s = JSON.parse(require('fs').readFileSync('$PREFS_SCHEMA', 'utf8'));
   const km = s.properties.global.properties.keychainMapping.properties;
   console.log(Object.keys(km).length);
 " 2>/dev/null)
-EXPECTED_SERVICES=13  # 12 + legacy "bitbucket" alias
-if [ "$KM_COUNT" = "$EXPECTED_SERVICES" ] || [ "$KM_COUNT" = "12" ]; then
-  pass "keychainMapping has $KM_COUNT services (≥12 expected)"
+if [ "${KM_COUNT:-0}" -ge 12 ] 2>/dev/null; then
+  pass "keychainMapping declares $KM_COUNT services (≥12 expected)"
 else
-  fail "keychainMapping has $KM_COUNT services, expected ≥12"
+  fail "keychainMapping declares ${KM_COUNT:-0} services, expected ≥12"
+fi
+
+MISSING_KEYS=$(node -e "
+  const s = JSON.parse(require('fs').readFileSync('$PREFS_SCHEMA', 'utf8'));
+  const km = Object.keys(s.properties.global.properties.keychainMapping.properties);
+  const required = ['jira','bitbucket_token','bitbucket_user','github','confluence','figma','figma_pat','figma_mcp','firebase','npm'];
+  console.log(required.filter((k) => !km.includes(k)).join(','));
+" 2>/dev/null)
+if [ -z "$MISSING_KEYS" ]; then
+  pass "every credential the pipeline resolves is declared"
+else
+  fail "keychainMapping missing declared-and-used key(s): $MISSING_KEYS"
 fi
 
 # ──────────────────────────────────────────────────────────────────────────

package/pipeline/scripts/smoke-shadow-git.sh

@@ -59,7 +59,9 @@ fi
 
 # 4-9. Lifecycle on a temp project
 TMP=$(mktemp -d)
-cd "$TMP"
+# A failed cd here would run the relative `rm -rf DerivedData` below in the
+# caller's directory instead of the sandbox.
+cd "$TMP" || { record_fail "cannot cd into sandbox $TMP"; exit 1; }
 git init --quiet
 echo "node_modules/" > .gitignore
 echo "hello" > a.txt
@@ -178,7 +180,7 @@ else
   record_fail "~/.claude/state is GONE after unsafe prune attempts"
 fi
 
-cd - >/dev/null
+cd - >/dev/null || true
 rm -rf "$TMP"
 
 # 10-12. Prefs schema

package/pipeline/scripts/triage-memory.mjs

@@ -63,9 +63,27 @@ function die(msg, code = 1) {
   process.exit(code);
 }
 
+// Resolves to the MAIN repository name even when called from inside a pipeline
+// worktree. `--show-toplevel` returns the worktree root, whose basename is the
+// task id, which would shard the per-repo store one directory per task and make
+// prior-art recall always miss. `--git-common-dir` points at the main
+// repository's .git for every linked worktree.
 function repoSlug() {
   if (flags["repo-slug"]) return String(flags["repo-slug"]);
   if (process.env.MULTI_AGENT_REPO_SLUG) return process.env.MULTI_AGENT_REPO_SLUG;
+  try {
+    const commonDir = execFileSync(
+      "git",
+      ["rev-parse", "--path-format=absolute", "--git-common-dir"],
+      { encoding: "utf8", stdio: ["ignore", "pipe", "ignore"] }
+    ).trim();
+    if (commonDir) {
+      const base = basename(commonDir);
+      const root = base === ".git" ? dirname(commonDir) : commonDir.replace(/\.git$/, "");
+      const slug = basename(root);
+      if (slug) return slug;
+    }
+  } catch { /* not a git repo, or git too old for --path-format */ }
   try {
     const top = execFileSync("git", ["rev-parse", "--show-toplevel"], {
       encoding: "utf8",
@@ -84,6 +102,33 @@ function ensureDir(p) {
   mkdirSync(dirname(p), { recursive: true });
 }
 
+// Freshness gate (P1): stamp each finding with the git blob SHA of its file at
+// ingest time; on query, if the file's current content hash differs, the hit is
+// flagged stale so a reused finding is verified, not trusted blindly.
+function repoTop() {
+  try {
+    return execFileSync("git", ["rev-parse", "--show-toplevel"], {
+      encoding: "utf8",
+      stdio: ["ignore", "pipe", "ignore"],
+    }).trim() || null;
+  } catch {
+    return null;
+  }
+}
+
+function fileSha(file, top) {
+  if (!file || !top) return null;
+  try {
+    return execFileSync("git", ["hash-object", "--", file], {
+      cwd: top,
+      encoding: "utf8",
+      stdio: ["ignore", "pipe", "ignore"],
+    }).trim() || null;
+  } catch {
+    return null;
+  }
+}
+
 const STOP = new Set([
   "the","a","an","and","or","but","of","in","on","at","to","for","with","by","is","are",
   "be","been","being","this","that","these","those","it","its","as","not","no","do","does",
@@ -157,6 +202,7 @@ function ingest() {
   catch (e) { die(`cannot parse triage JSON: ${e.message}`, 1); }
 
   const ts = new Date().toISOString();
+  const top = repoTop();
   const corpus = readCorpus(slug);
   const seen = new Set(corpus.map((r) => `${r.task_id}::${r.classification}::${r.file}::${r.issue}`));
 
@@ -188,6 +234,7 @@ function ingest() {
         fix: f.fix || null,
         reason: f._reason || null,
         reviewer: f.reviewer || null,
+        file_sha: fileSha(file, top),
       };
       writeRow(slug, row);
       seen.add(key);
@@ -212,21 +259,30 @@ function query() {
   const qTokens = tokenize(issueText);
   const fileRe = fileGlob ? globToRegExp(fileGlob) : null;
   const filtered = fileRe ? corpus.filter((r) => fileRe.test(r.file || "")) : corpus.slice();
+  const top = repoTop();
   const ranked = filtered
     .map((r) => ({ row: r, score: score(qTokens, r) }))
     .filter((x) => x.score > 0 || qTokens.size === 0)
     .sort((a, b) => b.score - a.score)
     .slice(0, topN)
-    .map(({ row, score }) => ({
-      taskId: row.task_id,
-      classification: row.classification,
-      severity: row.severity,
-      file: row.file,
-      line: row.line,
-      issue: row.issue,
-      reason: row.reason,
-      score: Math.round(score * 1000) / 1000,
-    }));
+    .map(({ row, score }) => {
+      // Freshness: null when we cannot compare (no stored sha, or file gone /
+      // not in a git repo); true means the file changed since the finding was
+      // recorded, so verify before reusing.
+      const cur = row.file_sha ? fileSha(row.file, top) : null;
+      const stale = row.file_sha && cur ? cur !== row.file_sha : null;
+      return {
+        taskId: row.task_id,
+        classification: row.classification,
+        severity: row.severity,
+        file: row.file,
+        line: row.line,
+        issue: row.issue,
+        reason: row.reason,
+        stale,
+        score: Math.round(score * 1000) / 1000,
+      };
+    });
   process.stdout.write(JSON.stringify({ ok: true, slug, total: filtered.length, hits: ranked }) + "\n");
   process.exit(0);
 }

package/pipeline/scripts/uninstall.mjs

@@ -305,7 +305,7 @@ async function confirm(question) {
   });
 }
 
-async function main() {
+export async function main() {
   console.log("");
   console.log("  multi-agent-pipeline uninstaller");
   console.log("  ================================");

package/pipeline/skills/.skill-manifest.json

@@ -1,11 +1,11 @@
 {
   "schemaVersion": "1.0.0",
-  "generatedAt": "2026-07-20T07:34:26Z",
-  "skillCount": 189,
+  "generatedAt": "2026-07-25T10:41:38Z",
+  "skillCount": 193,
   "entries": [
     {
       "path": "shared/core/apple-archive-compliance/SKILL.md",
-      "sha256": "38cb4c257f7b2915d1784ec6f355b9a7f14bb32427e09c01872d3c8d6a039a71"
+      "sha256": "d3826905606b6ec7d86d3434b03bfdbbd52a424f7108e58ca12a8cb490f1e4ad"
     },
     {
       "path": "shared/core/google-play-compliance/SKILL.md",
@@ -35,6 +35,10 @@
       "path": "shared/core/multi-agent-create-jira/SKILL.md",
       "sha256": "40caf0489ad8322cd00d792a244e772d4ddb6ccfd7c3807729d6b17590e7f81c"
     },
+    {
+      "path": "shared/core/multi-agent-design-check/SKILL.md",
+      "sha256": "a004cf24661789d2a3595c5246dfebe00b76aaee6ece057b458748f502a8cd5c"
+    },
     {
       "path": "shared/core/multi-agent-dev-autopilot/SKILL.md",
       "sha256": "82b1cf4a4bbd84fcfef54fcdbcdefa27f2747df6d4fac5c86b42e861013c6d58"
@@ -59,13 +63,17 @@
       "path": "shared/core/multi-agent-finish/SKILL.md",
       "sha256": "3e8e430f7aee626be96129a9d0ae6c9ac5ef0dbf80cbe580dde2d7fb5fccfed1"
     },
+    {
+      "path": "shared/core/multi-agent-forget/SKILL.md",
+      "sha256": "68f6a92a00844ed34c336548888d3e623534c12b5e85afe5762e7a9c8d41aae4"
+    },
     {
       "path": "shared/core/multi-agent-garbage-collect/SKILL.md",
       "sha256": "24736f163e645e2bee7d552e555ae6e32cd3fc3d15c844813a75b908a58d52fd"
     },
     {
       "path": "shared/core/multi-agent-help/SKILL.md",
-      "sha256": "00772bab27906cdbdcae72d115c4e1f12ddd914fa7aa274923618a3a906e6caa"
+      "sha256": "330780854ab3b48f788e34f918ecad585bd46ffbb44f14fd38de7b48a4b568aa"
     },
     {
       "path": "shared/core/multi-agent-issue/SKILL.md",
@@ -109,7 +117,7 @@
     },
     {
       "path": "shared/core/multi-agent-refactor/SKILL.md",
-      "sha256": "f8a9495f6aebacba634672d6c9416046521cbdfa769746c9ff3a4a993482a85a"
+      "sha256": "60766ee3a6b974b3872417db5c40d1fc03bfddd99daf4aca11d7a59b57a41b57"
     },
     {
       "path": "shared/core/multi-agent-resume/SKILL.md",
@@ -127,6 +135,14 @@
       "path": "shared/core/multi-agent-review/SKILL.md",
       "sha256": "b1fd57d2b80ee0fe2490439d9f2688fcf415857c2aab60549c5456e27693b9a9"
     },
+    {
+      "path": "shared/core/multi-agent-routines/SKILL.md",
+      "sha256": "91d3bdb15da7addadd1afd0b170057daf3d3fc335da3f6e1feb4feff1af0881e"
+    },
+    {
+      "path": "shared/core/multi-agent-save/SKILL.md",
+      "sha256": "f9c6e2ba1dd1b224b80cabe35f4327b5bcc0b1e05868b378547bec616193ef1b"
+    },
     {
       "path": "shared/core/multi-agent-scan/SKILL.md",
       "sha256": "266e4e7c9b702bc3205b7d926455c5863e144cddb1e4b883aa81e570a362fd3f"
@@ -149,11 +165,11 @@
     },
     {
       "path": "shared/core/multi-agent-sync/SKILL.md",
-      "sha256": "de38486056c99a599d93ae5166a2071875331a7d77f1e34f66d8b72657a73c2f"
+      "sha256": "3a9f7d8caa47d41d1a4759ce6673bfb94a1ef9f62f7113a4cd3b1e9451464601"
     },
     {
       "path": "shared/core/multi-agent-test/SKILL.md",
-      "sha256": "f90c34f71ea6cd5cccbfa128597eaaa221a0ca5f62924eb9cfb6b1ed2d07de04"
+      "sha256": "2b662ecd9497149c4c2e87cc79e906cc1cab7ea4a7d0efe300b0b018559960c4"
     },
     {
       "path": "shared/core/multi-agent-uninstall/SKILL.md",
@@ -425,7 +441,7 @@
     },
     {
       "path": "shared/external/humanizer/SKILL.md",
-      "sha256": "7f9d49b90eba79bf7acd8a5d8d5ae091cbc970188b8ecac310cce04b82add42e"
+      "sha256": "2fc8fe79a5832b9e1316f515c73f0545023153e85aba4645ba42060a0afdd049"
     },
     {
       "path": "shared/external/ios-accessibility/SKILL.md",