@mmerterden/multi-agent-pipeline 8.6.0

package/pipeline/scripts/smoke-readme-counts.sh

@@ -0,0 +1,120 @@
+#!/usr/bin/env bash
+# smoke-readme-counts.sh — fail when README "Current at-a-glance" counts drift
+# from the actual filesystem. The table is hand-edited; without this gate it
+# silently goes stale every time a smoke / skill / schema is added or removed.
+#
+# Each row is checked independently. Compute the live value, parse the README
+# value, fail if they differ. Output the exact table line so the fix is a
+# one-line edit.
+#
+# Exit 0 = every count matches reality, 1 = at least one row drifted.
+
+set -uo pipefail
+
+ROOT="$(cd "$(dirname "$0")/../.." && pwd)"
+README="$ROOT/README.md"
+
+pass=0
+fail=0
+failures=()
+record_pass() { pass=$((pass + 1)); printf '  \033[0;32mPASS\033[0m %s\n' "$1"; }
+record_fail() { fail=$((fail + 1)); failures+=("$1"); printf '  \033[0;31mFAIL\033[0m %s\n' "$1"; }
+
+printf '→ smoke-readme-counts: README "at-a-glance" table matches filesystem\n'
+
+[ -f "$README" ] || { record_fail "README.md missing"; exit 1; }
+
+# Helper: extract the integer at the end of a markdown table row whose first
+# cell starts with the given prefix. Examples of expected rows:
+#   "| Smoke suites | 73 |"
+#   "| Total `SKILL.md` files across all groups | 195 |"
+parse_count() {
+  local prefix_pattern="$1"
+  # Use grep -E for the line match (more predictable than awk's match()),
+  # then strip everything that isn't the trailing integer cell.
+  local row
+  row=$(grep -E "$prefix_pattern" "$README" | head -1)
+  [ -z "$row" ] && { echo "PARSE_ERROR"; return; }
+  # The integer is the last `| <digits> |` cell on the row.
+  echo "$row" | grep -oE '\| *[0-9]+ *\|' | tail -1 | grep -oE '[0-9]+'
+}
+
+check_count() {
+  local label="$1" actual="$2" prefix_pattern="$3"
+  local declared
+  declared=$(parse_count "$prefix_pattern")
+  if [ -z "$declared" ] || [ "$declared" = "PARSE_ERROR" ]; then
+    record_fail "$label: README row missing or unparseable (pattern: $prefix_pattern)"
+    return
+  fi
+  if [ "$declared" = "$actual" ]; then
+    record_pass "$label: README $declared matches actual $actual"
+  else
+    record_fail "$label: README claims $declared but actual is $actual"
+  fi
+}
+
+# --- Compute live counts -----------------------------------------------------
+
+# Slash commands: every .md in commands/multi-agent/ that is NOT an internal
+# `_`-prefixed picker fragment.
+SLASH=$(ls "$ROOT"/pipeline/commands/multi-agent/*.md 2>/dev/null \
+  | xargs -n1 basename | grep -v '^_' | wc -l | tr -d ' ')
+
+# Copilot skills: skill dirs whose SKILL.md frontmatter is `name: multi-agent-*`
+# AND `user-invocable: true`. Mirrors the Claude Code slash-command surface.
+COPILOT=$(grep -lE '^name: multi-agent-' "$ROOT"/pipeline/skills/shared/core/*/SKILL.md 2>/dev/null \
+  | xargs grep -l '^user-invocable: true' 2>/dev/null | wc -l | tr -d ' ')
+
+# Figma skills: SKILL.md count under figma-{common,ios,android}.
+FIGMA=$(find "$ROOT"/pipeline/skills/figma-common "$ROOT"/pipeline/skills/figma-ios "$ROOT"/pipeline/skills/figma-android \
+  -name SKILL.md 2>/dev/null | wc -l | tr -d ' ')
+
+# External skill catalog: subdirectories of shared/external/.
+EXTERNAL=$(ls -d "$ROOT"/pipeline/skills/shared/external/*/ 2>/dev/null | wc -l | tr -d ' ')
+
+# Total SKILL.md files: every SKILL.md anywhere under pipeline/skills/.
+TOTAL_SKILLS=$(find "$ROOT"/pipeline/skills -name SKILL.md 2>/dev/null | wc -l | tr -d ' ')
+
+# Smoke suites: every smoke-*.sh in pipeline/scripts/.
+SMOKES=$(ls "$ROOT"/pipeline/scripts/smoke-*.sh 2>/dev/null | wc -l | tr -d ' ')
+
+# Eval-triage fixtures: subdirs in pipeline/eval/triage/ whose names start with
+# a digit (the numbered fixture pattern, e.g. `01-empty-findings`).
+EVAL_TRIAGE=$(ls -d "$ROOT"/pipeline/eval/triage/[0-9]* 2>/dev/null | wc -l | tr -d ' ')
+
+# JSON schemas (excluding token-budget.json which is a config, not a schema).
+SCHEMAS=$(ls "$ROOT"/pipeline/schemas/*.schema.json 2>/dev/null | wc -l | tr -d ' ')
+
+# Agent personas: every .md under pipeline/agents/.
+AGENTS=$(find "$ROOT"/pipeline/agents -name '*.md' 2>/dev/null | wc -l | tr -d ' ')
+
+# Store-compliance skills (apple-archive-compliance + google-play-compliance).
+COMPLIANCE=$(ls -d "$ROOT"/pipeline/skills/shared/core/apple-archive-compliance "$ROOT"/pipeline/skills/shared/core/google-play-compliance 2>/dev/null | wc -l | tr -d ' ')
+
+# Golden-task fixtures.
+GOLDEN=$(ls -d "$ROOT"/pipeline/eval/golden-tasks/[0-9]* 2>/dev/null | wc -l | tr -d ' ')
+
+# --- Verify each row ---------------------------------------------------------
+
+check_count "slash commands"        "$SLASH"        '^\| Slash commands \(colon-form'
+check_count "Copilot skills"        "$COPILOT"      '^\| Copilot skills \(dash-form'
+check_count "store-compliance"      "$COMPLIANCE"   '^\| Store-compliance skills'
+check_count "figma skills"          "$FIGMA"        '^\| Figma skills'
+check_count "external catalog"      "$EXTERNAL"     '^\| External skill catalog'
+check_count "total SKILL.md"        "$TOTAL_SKILLS" '^\| Total `SKILL\.md`'
+check_count "smoke suites"          "$SMOKES"       '^\| Smoke suites'
+check_count "golden-task fixtures"  "$GOLDEN"       '^\| Golden-task fixtures'
+check_count "eval-triage fixtures"  "$EVAL_TRIAGE"  '^\| Eval-triage fixtures'
+check_count "JSON schemas"          "$SCHEMAS"      '^\| JSON schemas'
+check_count "agent personas"        "$AGENTS"       '^\| Agent personas'
+
+# --- Summary -----------------------------------------------------------------
+total=$((pass + fail))
+printf '\n→ smoke-readme-counts: %d/%d passed\n' "$pass" "$total"
+if [ "$fail" -ne 0 ]; then
+  printf '\nFailures (one-line README fix per row):\n'
+  for f in "${failures[@]}"; do printf '  - %s\n' "$f"; done
+  exit 1
+fi
+exit 0

package/pipeline/scripts/smoke-repo-map.sh

@@ -0,0 +1,300 @@
+#!/usr/bin/env bash
+# smoke-repo-map.sh
+#
+# Verifies pipeline/scripts/repo-map.mjs:
+#   1. CLI parses --root, --budget, --top, --format, --include, --exclude
+#   2. Empty repo → empty output (exit 0, never crashes)
+#   3. Swift fixture extracts class/struct/protocol/func
+#   4. Kotlin fixture extracts class/fun
+#   5. TS fixture extracts exported class/function
+#   6. Python fixture extracts class/def
+#   7. Bash fixture extracts top-level functions
+#   8. TF-IDF rule: an identifier defined in many files does not dominate rank
+#   9. Token budget honored — output stops at budget
+#  10. Determinism — same input twice = byte-identical output
+#  11. JSON format emits valid JSON with score / declarations
+#  12. Word-boundary match: substring of an identifier is NOT counted
+#
+# Exit 0 = all pass, 1 = any failure.
+
+set -euo pipefail
+
+ROOT="$(cd "$(dirname "$0")/../.." && pwd)"
+SCRIPT="$ROOT/pipeline/scripts/repo-map.mjs"
+
+pass=0
+fail=0
+failures=()
+record_pass() { pass=$((pass + 1)); printf '  \033[0;32mPASS\033[0m %s\n' "$1"; }
+record_fail() { fail=$((fail + 1)); failures+=("$1"); printf '  \033[0;31mFAIL\033[0m %s\n' "$1"; }
+
+printf '→ smoke-repo-map: Aider-style token-budgeted repo map contract\n'
+
+# Sanity: node + script
+if ! command -v node >/dev/null 2>&1; then
+  echo "SKIP smoke-repo-map: node not installed" >&2
+  exit 0
+fi
+if [ ! -f "$SCRIPT" ]; then
+  echo "FAIL: $SCRIPT missing" >&2
+  exit 1
+fi
+
+# --- Fixture builder --------------------------------------------------------
+FIX=$(mktemp -d)
+trap 'rm -rf "$FIX"' EXIT
+
+mkdir -p "$FIX/swift" "$FIX/kotlin" "$FIX/ts" "$FIX/py" "$FIX/bash" "$FIX/shared"
+
+# Swift
+cat > "$FIX/swift/Order.swift" <<'EOF'
+import Foundation
+
+public final class OrderViewModel {
+    public func loadOrders() async throws -> [Order] { return [] }
+}
+
+public struct OrderConfiguration {
+    var pageSize: Int = 20
+}
+
+public protocol OrderRepository {
+    func fetch() async throws -> [Order]
+}
+EOF
+
+cat > "$FIX/swift/Product.swift" <<'EOF'
+public class ProductViewModel {
+    public func loadProducts() async throws { _ = OrderViewModel() }
+}
+EOF
+
+# Kotlin
+cat > "$FIX/kotlin/User.kt" <<'EOF'
+package com.example
+
+data class User(val id: String, val name: String)
+
+class UserViewModel(private val repo: UserRepository) {
+    fun loadUsers(): List<User> = repo.all()
+}
+
+interface UserRepository {
+    fun all(): List<User>
+}
+EOF
+
+# TypeScript
+cat > "$FIX/ts/api.ts" <<'EOF'
+export class APIClient {
+  async fetchUsers(): Promise<User[]> { return []; }
+}
+export interface User {
+  id: string;
+  name: string;
+}
+export function buildClient(): APIClient { return new APIClient(); }
+EOF
+
+# Python
+cat > "$FIX/py/server.py" <<'EOF'
+class Server:
+    def __init__(self): pass
+
+def make_server() -> Server:
+    return Server()
+EOF
+
+# Bash
+cat > "$FIX/bash/helpers.sh" <<'EOF'
+#!/usr/bin/env bash
+common_helper() {
+  echo "hi"
+}
+specific_helper_for_repo() {
+  common_helper
+}
+EOF
+
+# Two shared-name files to test the TF-IDF rule (same identifier defined twice)
+cat > "$FIX/shared/A.swift" <<'EOF'
+class CommonName { var x = 1 }
+EOF
+cat > "$FIX/shared/B.swift" <<'EOF'
+class CommonName { var y = 2 }
+EOF
+
+# A reference file that mentions OrderViewModel (rare) AND CommonName (split)
+cat > "$FIX/shared/refs.swift" <<'EOF'
+import Foundation
+// References: OrderViewModel (defined once), CommonName (defined twice → split credit)
+let a = OrderViewModel()
+let b = CommonName()
+let c = CommonName()
+EOF
+
+# --- 1: CLI args parse without crashing ------------------------------------
+if node "$SCRIPT" --root "$FIX" --budget 800 --top 5 --format md >/dev/null 2>&1; then
+  record_pass "CLI parses standard flags"
+else
+  record_fail "CLI failed on standard flags"
+fi
+
+# --- 2: empty repo behaviour -----------------------------------------------
+EMPTY=$(mktemp -d)
+out2=$(node "$SCRIPT" --root "$EMPTY" --budget 200 --top 5 --format md 2>&1)
+rc=$?
+rm -rf "$EMPTY"
+if [ "$rc" -eq 0 ]; then
+  record_pass "empty repo → exit 0"
+else
+  record_fail "empty repo should exit 0 (got $rc)"
+fi
+
+# --- 3-7: language extractors --------------------------------------------
+MD=$(node "$SCRIPT" --root "$FIX" --budget 5000 --top 50 --format md 2>&1)
+
+if grep -qF "OrderViewModel" <<<"$MD" && grep -qF "OrderConfiguration" <<<"$MD" && grep -qF "OrderRepository" <<<"$MD"; then
+  record_pass "Swift extractor finds class/struct/protocol"
+else
+  record_fail "Swift extractor missed declarations"
+fi
+
+if grep -qF "UserViewModel" <<<"$MD" && grep -qF "UserRepository" <<<"$MD" && grep -qF "User" <<<"$MD"; then
+  record_pass "Kotlin extractor finds class/data class/interface"
+else
+  record_fail "Kotlin extractor missed declarations"
+fi
+
+if grep -qF "APIClient" <<<"$MD" && grep -qF "buildClient" <<<"$MD"; then
+  record_pass "TS extractor finds export class/function"
+else
+  record_fail "TS extractor missed declarations"
+fi
+
+if grep -qF "class Server" <<<"$MD" && grep -qF "def make_server" <<<"$MD"; then
+  record_pass "Python extractor finds class/def"
+else
+  record_fail "Python extractor missed declarations"
+fi
+
+if grep -qF "common_helper()" <<<"$MD" && grep -qF "specific_helper_for_repo()" <<<"$MD"; then
+  record_pass "Bash extractor finds functions"
+else
+  record_fail "Bash extractor missed functions"
+fi
+
+# --- 8: TF-IDF rule — rare identifier ranks higher than shared identifier ---
+# OrderViewModel defined once, referenced from refs.swift.
+# CommonName defined twice (A + B), referenced from refs.swift.
+# refs.swift's outgoing edge to Order.swift should weigh more than to A/B
+# combined — so Order.swift should rank above A.swift in this fixture.
+order_line=$(echo "$MD" | grep -nF "Order.swift" | head -1 | cut -d: -f1)
+a_line=$(echo "$MD" | grep -nF "shared/A.swift" | head -1 | cut -d: -f1)
+if [ -n "$order_line" ] && [ -n "$a_line" ]; then
+  if [ "$order_line" -lt "$a_line" ]; then
+    record_pass "TF-IDF — rarely-defined identifier outranks shared identifier"
+  else
+    record_fail "TF-IDF rule failed: Order.swift should rank above shared/A.swift"
+  fi
+else
+  record_fail "TF-IDF — could not locate Order.swift or shared/A.swift in output"
+fi
+
+# --- 9: token budget honored -----------------------------------------------
+SMALL=$(node "$SCRIPT" --root "$FIX" --budget 150 --top 50 --format md 2>&1)
+small_bytes=$(printf '%s' "$SMALL" | wc -c)
+# 150 tokens × 4 chars/token = 600 chars; output may end on file boundary so allow +200
+if [ "$small_bytes" -lt 900 ]; then
+  record_pass "token budget honored (budget=150 → $small_bytes chars)"
+else
+  record_fail "token budget exceeded ($small_bytes chars for budget=150)"
+fi
+
+# --- 10: determinism --------------------------------------------------------
+RUN1=$(node "$SCRIPT" --root "$FIX" --budget 800 --top 10 --format md 2>&1)
+RUN2=$(node "$SCRIPT" --root "$FIX" --budget 800 --top 10 --format md 2>&1)
+if [ "$RUN1" = "$RUN2" ]; then
+  record_pass "determinism — same input twice produces byte-identical output"
+else
+  record_fail "non-deterministic output across runs"
+fi
+
+# --- 11: JSON format --------------------------------------------------------
+JSON=$(node "$SCRIPT" --root "$FIX" --budget 2000 --top 5 --format json 2>&1)
+if echo "$JSON" | python3 -c 'import json,sys; data=json.load(sys.stdin); assert isinstance(data,list); assert all("path" in d and "score" in d and "declarations" in d for d in data); print("ok")' >/dev/null 2>&1; then
+  record_pass "JSON format produces valid array with score+declarations"
+else
+  record_fail "JSON output failed schema check"
+fi
+
+# --- 12: word-boundary match -----------------------------------------------
+# Isolated fixture: define a unique identifier in one file, reference its
+# *substring* (no word boundary) in another, then verify in_refs stays 0.
+WB=$(mktemp -d)
+cat > "$WB/def.swift" <<'EOF'
+class WidgetXYZ123 { var x = 0 }
+EOF
+cat > "$WB/ref.swift" <<'EOF'
+// Mentions a SUPER-string only: WidgetXYZ123Helper.
+// \bWidgetXYZ123\b should NOT match inside WidgetXYZ123Helper.
+struct WidgetXYZ123Helper { }
+EOF
+WB_JSON=$(node "$SCRIPT" --root "$WB" --budget 2000 --top 5 --format json 2>&1)
+rm -rf "$WB"
+
+def_refs=$(echo "$WB_JSON" | python3 -c '
+import json, sys
+data = json.load(sys.stdin)
+for d in data:
+    if d["path"] == "def.swift":
+        print(d["incoming_refs"])
+        break
+')
+if [ -n "$def_refs" ] && python3 -c "import sys; sys.exit(0 if abs(${def_refs}) < 0.01 else 1)" 2>/dev/null; then
+  record_pass "word-boundary — super-string match (WidgetXYZ123Helper) does not count as WidgetXYZ123 ref"
+else
+  record_fail "word-boundary regression: def.swift in_refs=$def_refs (expected ~0)"
+fi
+
+# --- 13: prefs schema exposes repoMap toggle -------------------------------
+SCHEMA_PATH="$ROOT/pipeline/schemas/prefs.schema.json"
+if jq -e '.properties.global.properties.repoMap.properties.enabled' "$SCHEMA_PATH" >/dev/null 2>&1; then
+  record_pass "prefs schema exposes repoMap.enabled"
+else
+  record_fail "prefs schema missing repoMap.enabled"
+fi
+for prop in tokenBudget topFiles include exclude; do
+  if jq -e ".properties.global.properties.repoMap.properties.${prop}" "$SCHEMA_PATH" >/dev/null 2>&1; then
+    record_pass "prefs schema exposes repoMap.${prop}"
+  else
+    record_fail "prefs schema missing repoMap.${prop}"
+  fi
+done
+# Off by default — preserve baseline for existing users.
+if jq -e '.properties.global.properties.repoMap.properties.enabled
+          | has("default") and .default == false' "$SCHEMA_PATH" >/dev/null 2>&1; then
+  record_pass "repoMap.enabled defaults to false (opt-in)"
+else
+  record_fail "repoMap.enabled should default to false"
+fi
+
+# --- 14: Phase 1 + Phase 4 docs wire the injection ------------------------
+P1="$ROOT/pipeline/commands/multi-agent/refs/phases/phase-1-analysis.md"
+P4="$ROOT/pipeline/commands/multi-agent/refs/phases/phase-4-review.md"
+for f in "$P1" "$P4"; do
+  base=$(basename "$f")
+  if grep -qF "repoMap.enabled" "$f" && grep -qF "repo-map.mjs" "$f" && grep -qF '${REPO_MAP}' "$f"; then
+    record_pass "$base wires repoMap injection"
+  else
+    record_fail "$base missing repoMap wiring (need: repoMap.enabled + repo-map.mjs + \${REPO_MAP})"
+  fi
+done
+
+printf '\n══ repo-map smoke: %d passed, %d failed ══\n' "$pass" "$fail"
+if [ "$fail" -gt 0 ]; then
+  printf '\nFailures:\n'
+  for msg in "${failures[@]}"; do printf '  - %s\n' "$msg"; done
+  exit 1
+fi
+exit 0

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

@@ -0,0 +1,134 @@
+#!/usr/bin/env bash
+# smoke-review-watch.sh
+#
+# Verifies the auto-review-incoming-PRs surface:
+#   1. pipeline/lib/review-watch.sh exists and parses cleanly
+#   2. --doctor runs without crashing on an unconfigured environment
+#   3. --help shows usage
+#   4. State directory is created
+#   5. prefs schema exposes global.reviewWatch.{enabled,repos,intervalSeconds,labelFilter}
+#   6. global.reviewWatch.enabled defaults to false (opt-in)
+#   7. global.review.dedupeInlineComments default is true (Bugbot parity)
+#   8. post-pr-review.sh exposes finding_fingerprint + comment_exists_with_fingerprint
+#   9. render_inline_body accepts a 5th fingerprint arg and embeds the marker
+#  10. Unknown command exits non-zero
+#
+# Exit 0 = all pass, 1 = any failure.
+
+set -uo pipefail
+
+ROOT="$(cd "$(dirname "$0")/../.." && pwd)"
+WATCH="$ROOT/pipeline/lib/review-watch.sh"
+POSTER="$ROOT/pipeline/lib/post-pr-review.sh"
+SCHEMA="$ROOT/pipeline/schemas/prefs.schema.json"
+
+pass=0
+fail=0
+failures=()
+record_pass() { pass=$((pass + 1)); printf '  \033[0;32mPASS\033[0m %s\n' "$1"; }
+record_fail() { fail=$((fail + 1)); failures+=("$1"); printf '  \033[0;31mFAIL\033[0m %s\n' "$1"; }
+
+printf '→ smoke-review-watch: incoming-PR auto-review surface contract\n'
+
+# 1. lib script exists + parses
+if [ ! -f "$WATCH" ]; then
+  record_fail "pipeline/lib/review-watch.sh missing"
+elif ! bash -n "$WATCH" 2>/dev/null; then
+  record_fail "review-watch.sh syntax error"
+else
+  record_pass "review-watch.sh exists and parses"
+fi
+
+# 2. --doctor runs cleanly (zero exit) on unconfigured prefs
+if bash "$WATCH" --doctor >/dev/null 2>&1; then
+  record_pass "--doctor exits 0 on unconfigured environment"
+else
+  record_fail "--doctor crashed (must be safe on unconfigured environment)"
+fi
+
+# 3. --help shows usage
+if bash "$WATCH" --help 2>&1 | grep -qE 'review-watch|usage:'; then
+  record_pass "--help shows usage"
+else
+  record_fail "--help should show usage banner"
+fi
+
+# 4. State dir is created
+STATE_DIR="$HOME/.claude/state/review-watch"
+if [ -d "$STATE_DIR" ]; then
+  record_pass "state dir created at $STATE_DIR"
+else
+  record_fail "state dir not created"
+fi
+
+# 5. Schema toggles
+for prop in enabled repos intervalSeconds labelFilter; do
+  if jq -e ".properties.global.properties.reviewWatch.properties.${prop}" "$SCHEMA" >/dev/null 2>&1; then
+    record_pass "prefs schema exposes reviewWatch.${prop}"
+  else
+    record_fail "prefs schema missing reviewWatch.${prop}"
+  fi
+done
+
+# 6. reviewWatch.enabled defaults to false
+if jq -e '.properties.global.properties.reviewWatch.properties.enabled
+          | has("default") and .default == false' "$SCHEMA" >/dev/null 2>&1; then
+  record_pass "reviewWatch.enabled defaults to false (opt-in)"
+else
+  record_fail "reviewWatch.enabled should default to false"
+fi
+
+# 7. dedupeInlineComments default true
+if jq -e '.properties.global.properties.review.properties.dedupeInlineComments
+          | has("default") and .default == true' "$SCHEMA" >/dev/null 2>&1; then
+  record_pass "review.dedupeInlineComments defaults to true (Bugbot parity)"
+else
+  record_fail "review.dedupeInlineComments should default to true"
+fi
+
+# 8. post-pr-review.sh exposes the dedupe helpers (provider-agnostic)
+if grep -qE '^finding_fingerprint\(\)' "$POSTER" \
+   && grep -qE '^comment_exists_with_fingerprint\(\)' "$POSTER" \
+   && grep -qE '^load_existing_comments_github\(\)' "$POSTER" \
+   && grep -qE '^load_existing_comments_bitbucket\(\)' "$POSTER"; then
+  record_pass "post-pr-review.sh declares dedupe helpers (fingerprint + check + github + bitbucket loaders)"
+else
+  record_fail "post-pr-review.sh missing one of the dedupe helpers (github or bitbucket loader)"
+fi
+
+# 8b. Bitbucket dedupe gate wired in post_bitbucket_server
+if awk '/^post_bitbucket_server\(\)/,/^}/' "$POSTER" \
+   | grep -q 'load_existing_comments_bitbucket'; then
+  record_pass "post_bitbucket_server wires the dedupe gate"
+else
+  record_fail "post_bitbucket_server missing load_existing_comments_bitbucket call"
+fi
+if awk '/^post_bitbucket_server\(\)/,/^}/' "$POSTER" \
+   | grep -q 'comment_exists_with_fingerprint'; then
+  record_pass "post_bitbucket_server checks fingerprint before posting"
+else
+  record_fail "post_bitbucket_server missing fingerprint check"
+fi
+
+# 9. render_inline_body takes 5 args (sev issue fix rule_id fingerprint) and writes the marker
+if grep -qE 'render_inline_body\(\)' "$POSTER" \
+   && grep -qE '<!-- multi-agent-finding:' "$POSTER"; then
+  record_pass "render_inline_body embeds the fingerprint marker"
+else
+  record_fail "render_inline_body must embed <!-- multi-agent-finding: ... --> marker"
+fi
+
+# 10. Unknown command exits non-zero
+if bash "$WATCH" --garbage >/dev/null 2>&1; then
+  record_fail "unknown command should exit non-zero"
+else
+  record_pass "unknown command exits non-zero"
+fi
+
+printf '\n══ review-watch smoke: %d passed, %d failed ══\n' "$pass" "$fail"
+if [ "$fail" -gt 0 ]; then
+  printf '\nFailures:\n'
+  for msg in "${failures[@]}"; do printf '  - %s\n' "$msg"; done
+  exit 1
+fi
+exit 0