ci(cold-db-smoke): inline into workflow; remove the script (operator: not a per-commit gate)

Operator pushback: 'I don't want a smoke test I have to manually run
every time I commit.' Correct read — the script existed for local
debugging but its presence in scripts/ci-guards/ implied 'operator
runs this regularly,' which is the opposite of the design intent.

Changes:

- Removed scripts/ci-guards/cold-db-compose-smoke.sh.
- Inlined the smoke logic directly into the
  cold-db-compose-smoke job in .github/workflows/ci.yml. Same
  semantics: docker compose down -v -> up -d -> wait-healthy ->
  bootstrap admin -> issue/renew/revoke -> assert audit rows ->
  teardown. 15-min wall-clock cap. Logs dump on failure.
- Removed the cold-db-compose-smoke.sh skip case from the generic
  regression-guards loop (no longer needed).
- Updated scripts/ci-guards/README.md and
  docs/contributor/ci-guards.md to reflect the new shape: 'lives in
  the workflow, not as a script.'

Workspace docs updated (cowork/WORKSPACE-CHANGELOG.md,
cowork/CLAUDE.md, cowork/auditable-codebase-bundle/RESULTS.md).

The gate is unchanged: CI runs the smoke on every push, master
branch-protection enforces it as a required check. Operator's
manual action is once — adding the check to branch-protection.

Audit-Closes: post-v2.1.0-anti-rot/item-6

.github/workflows/ci.yml

@@ -210,22 +210,10 @@ jobs:
         # Contract: each guard MUST exit 0 on clean repo, non-zero with
         # ::error:: prefix on regression. See scripts/ci-guards/README.md.
         #
-        # SKIP cold-db-compose-smoke.sh — it needs Docker + a fresh
-        # postgres volume, which only exists in the dedicated
-        # `cold-db-compose-smoke` job below. Including it in this loop
-        # would always fail (no Docker on the runners that don't bring
-        # up compose).
         run: |
           set -e
           fail=0
           for g in scripts/ci-guards/*.sh; do
-            case "$(basename "$g")" in
-              cold-db-compose-smoke.sh)
-                echo "::group::$(basename "$g") (skipped — runs in dedicated job)"
-                echo "::endgroup::"
-                continue
-                ;;
-            esac
             echo "::group::$(basename "$g")"
             if ! bash "$g"; then
               fail=1
@@ -258,13 +246,108 @@ jobs:
         # 15-min wall-clock cap covers cold image pull + compose-up +
         # full issue/renew/revoke probe + teardown. Increase only if
         # the underlying steps legitimately grow.
+        #
+        # The smoke is inlined here on purpose — it is NOT a script in
+        # scripts/ci-guards/, because there is no value in a developer
+        # running this locally. The whole point of the gate is that CI
+        # owns the cold-DB state; the operator never has to remember to
+        # run it. Master branch-protection enforces this job as a
+        # required check; that is the manual action, and it happens
+        # once.
         timeout-minutes: 15
-        run: bash scripts/ci-guards/cold-db-compose-smoke.sh
+        working-directory: deploy
+        env:
+          STARTUP_TIMEOUT_SECONDS: 300
+        run: |
+          set -e
+          set -o pipefail
+
+          SERVER_URL="https://localhost:8443"
+          CACERT_PATH="${GITHUB_WORKSPACE}/deploy/test/certs/ca.crt"
+
+          log() { echo "[cold-db-smoke] $*"; }
+
+          wait_for_service_healthy() {
+            local svc="$1" deadline=$(( $(date +%s) + STARTUP_TIMEOUT_SECONDS ))
+            while [ "$(date +%s)" -lt "$deadline" ]; do
+              local state
+              state="$(docker compose ps --format json "$svc" 2>/dev/null | python3 -c '
+          import json, sys
+          try:
+              line = sys.stdin.read().strip()
+              if not line:
+                  print("not-up"); sys.exit(0)
+              rows = json.loads(line) if line.startswith("[") else [json.loads(l) for l in line.splitlines() if l.strip()]
+              if not rows:
+                  print("not-up")
+              else:
+                  print(rows[0].get("Health", rows[0].get("State", "?")))
+          except Exception as e:
+              print(f"err: {e}")
+          ')"
+              if [ "$state" = "healthy" ] || [ "$state" = "running" ]; then
+                log "  $svc → $state"; return 0
+              fi
+              sleep 2
+            done
+            log "  $svc did NOT reach healthy within ${STARTUP_TIMEOUT_SECONDS}s (last: $state)"
+            return 1
+          }
+
+          http_call() {
+            local method="$1" path="$2" data="${3:-}"
+            local args=(--silent --show-error --max-time 30 -X "$method" "$SERVER_URL$path")
+            [ -f "$CACERT_PATH" ] && args+=(--cacert "$CACERT_PATH") || args+=(--insecure)
+            [ -n "${KEY:-}" ] && args+=(-H "Authorization: Bearer $KEY")
+            [ -n "$data" ] && args+=(-H "Content-Type: application/json" -d "$data")
+            curl "${args[@]}"
+          }
+
+          log "1/7 down -v --remove-orphans"
+          docker compose down -v --remove-orphans 2>&1 | tail -3 || true
+
+          log "2/7 up -d (cold boot)"
+          docker compose up -d 2>&1 | tail -3
+
+          log "3/7 wait for healthchecks"
+          wait_for_service_healthy postgres
+          wait_for_service_healthy certctl-server
+          wait_for_service_healthy certctl-agent || log "  (agent skipped — non-demo compose)"
+
+          log "4/7 minting day-0 admin"
+          TOKEN="$(openssl rand -base64 32 | tr -d '\n')"
+          echo "CERTCTL_BOOTSTRAP_TOKEN=$TOKEN" > /tmp/_smoke.env
+          docker compose --env-file /tmp/_smoke.env up -d --force-recreate certctl-server 2>&1 | tail -2
+          sleep 5
+          wait_for_service_healthy certctl-server
+          BODY="$(http_call POST /api/v1/auth/bootstrap "{\"token\":\"$TOKEN\",\"actor_name\":\"smoke-admin\"}")"
+          KEY="$(echo "$BODY" | python3 -c 'import json,sys; print(json.load(sys.stdin)["key_value"])')"
+          [ -n "$KEY" ] || { log "bootstrap failed: $BODY"; exit 1; }
+
+          log "5/7 issuing test cert"
+          ISSUE='{"common_name":"smoke-test.local","profile_id":"profile-default","environment":"test","owner_id":"o-platform"}'
+          R="$(http_call POST /api/v1/certificates "$ISSUE")"
+          CID="$(echo "$R" | python3 -c 'import json,sys; d=json.load(sys.stdin); print(d.get("id") or d.get("certificate",{}).get("id",""))')"
+          [ -n "$CID" ] || { log "issue failed: $R"; exit 1; }
+
+          log "6/7 renewing $CID"
+          http_call POST "/api/v1/certificates/$CID/renew" >/dev/null
+
+          log "7/7 revoking + asserting audit rows"
+          http_call POST "/api/v1/certificates/$CID/revoke" '{"reason":"smoke-test"}' >/dev/null
+          AUD="$(http_call GET '/api/v1/audit?limit=50')"
+          for action in cert.issued cert.renewed cert.revoked; do
+            if ! echo "$AUD" | python3 -c "import json,sys; d=json.load(sys.stdin); evs=d.get('events') or d.get('audit',{}).get('events') or []; sys.exit(0 if any(e.get('action')=='$action' for e in evs) else 1)"; then
+              log "MISSING audit row: $action"; echo "$AUD" | head -200; exit 1
+            fi
+          done
+          log "PASS — tearing down"
+          docker compose down -v 2>&1 | tail -2
 
       - name: Dump compose logs on failure
         if: failure()
+        working-directory: deploy
         run: |
-          cd deploy
           for svc in postgres certctl-server certctl-agent certctl-tls-init; do
             echo "==== $svc ===="
             docker compose logs --no-color --tail 200 "$svc" || true

docs/contributor/ci-guards.md

@@ -32,7 +32,9 @@ Catch drift across the four product surfaces — OpenAPI spec, HTTP router, MCP
 
 ### Operational guards
 
-`cold-db-compose-smoke.sh` (wipe postgres volume, bring stack up cold, issue/renew/revoke, audit-row check). `doc-rot-detector.sh` (every doc reviewed within 120 days). These pin the operational reality, not the source shape.
+`doc-rot-detector.sh` (every doc reviewed within 120 days) pins the operational reality, not the source shape.
+
+The cold-DB compose smoke (wipe postgres volume, bring stack up cold, issue/renew/revoke, audit-row check) lives directly in `.github/workflows/ci.yml::cold-db-compose-smoke` — not as a script. It is intentionally not operator-runnable: the gate's value is that CI owns the cold-DB state, the operator never has to remember to run it. Master branch-protection enforces the job as a required check; that is the manual action, and it happens once.
 
 ## When the build is red
 

scripts/ci-guards/README.md

@@ -92,7 +92,8 @@ These guards catch defect classes BEFORE they get audit findings — they pin in
 |---|---|---|
 | `complete-path-config-coverage` | post-v2.1.0 / item-1 | "Lying field" — `CERTCTL_*` env var defined in `internal/config/config.go` that no consumer outside `internal/config/` actually reads. Operator-facing config that the docs claim works but the code never honors. Companion Go test at `internal/config/coverage_test.go`. |
 | `doc-rot-detector` | post-v2.1.0 / item-5 | Docs older than 90 days warn (yellow), older than 120 days fail (red). Uses HEAD commit timestamp for reproducibility. `docs/archive/` allowlisted in bulk. |
-| `cold-db-compose-smoke` | post-v2.1.0 / item-6 | Migration-on-cold-DB regression (canonical: 2026-05-09 migration 000045 broken INSERT, commit `6444e13`). Wipes the postgres volume, brings the stack up cold, issue/renew/revoke + 3 audit rows. **Runs in its own GitHub Actions job** (`cold-db-compose-smoke`), NOT the generic regression-guards loop — needs Docker. |
+
+The cold-DB compose smoke (post-v2.1.0 / item-6) is NOT a script in this directory — it is inlined directly into `.github/workflows/ci.yml::cold-db-compose-smoke` because there is no value in a developer running it locally (the whole point of the gate is that CI owns the cold-DB state). To inspect or modify the smoke logic, read that workflow job; there is intentionally no `scripts/ci-guards/cold-db-compose-smoke.sh`.
 
 The fourth Bundle artifact (`internal/ciparity/`) is Go tests, not shell guards — runs under the standard Go test step. Pins the MCP tool catalogue floor + naming convention; reports CLI/MCP/OpenAPI surface counts as a trend metric.
 

scripts/ci-guards/cold-db-compose-smoke.sh

@@ -1,180 +0,0 @@
-#!/usr/bin/env bash
-# scripts/ci-guards/cold-db-compose-smoke.sh
-#
-# Per post-v2.1.0 anti-rot item 6 (Auditable Codebase Bundle).
-#
-# The bug class this catches: a migration whose .up.sql is broken in a
-# way the unit tests / integration suite misses because they reuse a
-# warm DB across runs. The canonical case: 2026-05-09 migration
-# 000045's broken INSERT, surfaced only by a cold `docker compose up`
-# and fixed in commit 6444e13. This guard runs that very check on
-# every push.
-#
-# Workflow:
-#   1. docker compose down -v --remove-orphans         (wipe volumes)
-#   2. docker compose up -d                            (cold boot)
-#   3. wait up to 5 min for healthchecks               (postgres,
-#                                                       certctl-server,
-#                                                       certctl-agent)
-#   4. mint a day-0 admin via /api/v1/auth/bootstrap   (Bundle 1 path)
-#   5. issue + renew + revoke a certificate            (HTTP API)
-#   6. assert audit rows for each step
-#   7. docker compose down -v                          (clean up)
-#
-# Total runtime: ~3-5 min on warm Docker, ~5-10 min cold.
-#
-# Failure paths dump `docker compose logs` for every service to make
-# CI failures actionable without a re-run.
-#
-# This script is invoked by .github/workflows/ci.yml::cold-db-compose-smoke.
-# Runs locally for developers via `bash scripts/ci-guards/cold-db-compose-smoke.sh`.
-
-set -e
-set -o pipefail
-
-REPO_ROOT="$(cd "$(dirname "$0")/../.." && pwd)"
-cd "$REPO_ROOT/deploy"
-
-# Tunables (the CI job overrides these as needed).
-STARTUP_TIMEOUT_SECONDS="${COLD_DB_SMOKE_STARTUP_TIMEOUT:-300}"   # 5 min
-PROBE_TIMEOUT_SECONDS="${COLD_DB_SMOKE_PROBE_TIMEOUT:-180}"       # 3 min
-SERVER_URL="${COLD_DB_SMOKE_SERVER_URL:-https://localhost:8443}"
-CACERT_PATH="${COLD_DB_SMOKE_CACERT:-${REPO_ROOT}/deploy/test/certs/ca.crt}"
-
-# --- helpers ----------------------------------------------------------------
-
-log() { echo "[cold-db-smoke] $*"; }
-
-dump_logs_on_failure() {
-  log "FAILURE — dumping service logs:"
-  docker compose ps || true
-  for svc in postgres certctl-server certctl-agent certctl-tls-init; do
-    echo
-    echo "==== $svc ===="
-    docker compose logs --no-color --tail 200 "$svc" 2>&1 || true
-  done
-}
-
-trap 'dump_logs_on_failure' ERR
-
-wait_for_service_healthy() {
-  local svc="$1" deadline=$(( $(date +%s) + STARTUP_TIMEOUT_SECONDS ))
-  while [ "$(date +%s)" -lt "$deadline" ]; do
-    local state
-    state="$(docker compose ps --format json "$svc" 2>/dev/null | python3 -c '
-import json, sys
-try:
-    line = sys.stdin.read().strip()
-    if not line:
-        print("not-up")
-        sys.exit(0)
-    # docker compose ps emits one JSON object per line (NDJSON) on newer
-    # versions; older versions emit a JSON array. Handle both.
-    if line.startswith("["):
-        rows = json.loads(line)
-    else:
-        rows = [json.loads(l) for l in line.splitlines() if l.strip()]
-    if not rows:
-        print("not-up")
-    else:
-        print(rows[0].get("Health", rows[0].get("State", "?")))
-except Exception as e:
-    print(f"err: {e}")
-')"
-    if [ "$state" = "healthy" ] || [ "$state" = "running" ]; then
-      log "  $svc → $state"
-      return 0
-    fi
-    sleep 2
-  done
-  log "  $svc did NOT reach healthy within $STARTUP_TIMEOUT_SECONDS s (last state: $state)"
-  return 1
-}
-
-http_call() {
-  # http_call <method> <path> [data_json]
-  local method="$1" path="$2" data="${3:-}"
-  local args=(--silent --show-error --max-time 30 -X "$method" "$SERVER_URL$path")
-  if [ -f "$CACERT_PATH" ]; then
-    args+=(--cacert "$CACERT_PATH")
-  else
-    args+=(--insecure)
-  fi
-  if [ -n "${KEY:-}" ]; then
-    args+=(-H "Authorization: Bearer $KEY")
-  fi
-  if [ -n "$data" ]; then
-    args+=(-H "Content-Type: application/json" -d "$data")
-  fi
-  curl "${args[@]}"
-}
-
-# --- the smoke ---------------------------------------------------------------
-
-log "1/7 down -v --remove-orphans (wiping postgres volume)"
-docker compose down -v --remove-orphans 2>&1 | tail -3 || true
-
-log "2/7 up -d (cold boot)"
-docker compose up -d 2>&1 | tail -3
-
-log "3/7 waiting for healthchecks (timeout ${STARTUP_TIMEOUT_SECONDS}s/svc)"
-wait_for_service_healthy postgres
-wait_for_service_healthy certctl-server
-# certctl-agent depends on the demo seed having run; only wait when
-# CERTCTL_DEMO_SEED=true is in effect (the bundled clean compose
-# doesn't always seed). Best-effort.
-wait_for_service_healthy certctl-agent || log "  (agent healthcheck skipped — non-demo compose)"
-
-log "4/7 minting day-0 admin via /api/v1/auth/bootstrap"
-TOKEN="$(openssl rand -base64 32 | tr -d '\n')"
-# Restart the server with the bootstrap token so the strategy is
-# active. Compose --env-file is the lightest path.
-echo "CERTCTL_BOOTSTRAP_TOKEN=$TOKEN" > /tmp/_smoke.env
-docker compose --env-file /tmp/_smoke.env up -d --force-recreate certctl-server 2>&1 | tail -2
-sleep 5
-wait_for_service_healthy certctl-server
-
-BOOTSTRAP_BODY="$(http_call POST /api/v1/auth/bootstrap "{\"token\":\"$TOKEN\",\"actor_name\":\"smoke-admin\"}")"
-KEY="$(echo "$BOOTSTRAP_BODY" | python3 -c 'import json,sys; print(json.load(sys.stdin)["key_value"])')"
-if [ -z "$KEY" ]; then
-  log "  bootstrap did NOT return a key_value — body was:"
-  echo "$BOOTSTRAP_BODY"
-  exit 1
-fi
-log "  admin minted (actor=smoke-admin)"
-
-log "5/7 issuing a test certificate"
-# Use the default profile + demo CA. The exact shape may need a tweak
-# depending on the compose's seeded issuers — fail loudly with the
-# response body if the API rejects the request.
-ISSUE_BODY='{"common_name":"smoke-test.local","profile_id":"profile-default","environment":"test","owner_id":"o-platform"}'
-ISSUE_RESP="$(http_call POST /api/v1/certificates "$ISSUE_BODY")"
-CERT_ID="$(echo "$ISSUE_RESP" | python3 -c 'import json,sys; d=json.load(sys.stdin); print(d.get("id") or d.get("certificate",{}).get("id",""))')"
-if [ -z "$CERT_ID" ]; then
-  log "  issue failed; response body:"
-  echo "$ISSUE_RESP"
-  exit 1
-fi
-log "  cert issued: $CERT_ID"
-
-log "6/7 renewing the certificate"
-http_call POST "/api/v1/certificates/$CERT_ID/renew" >/dev/null
-log "  renewed"
-
-log "7/7 revoking the certificate + asserting audit rows"
-http_call POST "/api/v1/certificates/$CERT_ID/revoke" '{"reason":"smoke-test"}' >/dev/null
-AUDIT_BODY="$(http_call GET "/api/v1/audit?limit=50")"
-for action in cert.issued cert.renewed cert.revoked; do
-  if ! echo "$AUDIT_BODY" | python3 -c "import json,sys; d=json.load(sys.stdin); evs=d.get('events') or d.get('audit',{}).get('events') or []; sys.exit(0 if any(e.get('action')=='$action' for e in evs) else 1)"; then
-    log "  MISSING audit row: $action"
-    echo "$AUDIT_BODY" | head -200
-    exit 1
-  fi
-done
-log "  audit rows present: cert.issued, cert.renewed, cert.revoked"
-
-log "DONE — tearing down"
-trap - ERR
-docker compose down -v 2>&1 | tail -2
-rm -f /tmp/_smoke.env
-log "PASS"