feat(skills): wire verify+secure into ship-feature/init-project; onboard no-loop + LRN-093 backstop (verify-loops lot 5)

ship-feature: STEP 0e CONTRACT (request verbatim) → ENRICHED at the STEP 3
validation gate (design criteria appended [gated <date>], the human
micro-gate) → STEP 5 VERIFY+SECURE judges the branch against the ENRICHED
contract via the shared include. Distinct axis from STEP 6 code review, both
run (LRN-095).

init-project: contract seeded from the PROJECT BRIEF (V1 features → criteria)
→ ENRICHED at VALIDATION GATE #1 → STEP 9 VERIFY+SECURE. Adds the security
gate init-project previously lacked (was deferred to a later /onboard).

onboard: explicit NO verify-loop — it produces an audit report, not a change
to verify against a request; contract is scope-only, security-auditor runs
MODE audit (report-only), never a gate. Documented to prevent a misplaced
symmetry loop (BDR-050: dev pipeline != audit).

lib/tests/no-vacuous-locks.test.sh: deterministic backstop for LRN-093 (2nd
recurrence in this chantier → the advisory alone did not hold). Refuses a
literal \n in any grep/tf/tr_/tn pattern across lib/tests/*.test.sh;
flip-tested against a synthetic offender so the guard proves it bites.
lib/tests/loops-heavy.test.sh: 18 structure locks green.

Behavioral dogfood (both vigilance points, real): (1) enrichment — a fresh
verifier reads and checks a [gated] design criterion (ECARTS naming it
precisely); (2) escalation — 3 consecutive ECARTS on the same criterion →
orchestrator STOPs at the max-3 bound + presents the CONTRACT-vs-REALIZED
table, no 4th loop, no commit. First real exercise of the infinite-loop guard.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XpphkdTosUzokBDNG7PToS

lib/tests/loops-heavy.test.sh

@@ -0,0 +1,49 @@
+#!/usr/bin/env bash
+# ============================================================
+# Structure locks — heavy-flow wiring (verify-loops lot 5)
+# ship-feature + init-project get contract + enrich-at-gate +
+# verify-secure-loop; onboard is the explicit NO-LOOP audit case.
+# ============================================================
+set -u
+
+REPO="$(cd "$(dirname "$0")/../.." && pwd)"
+SHF="$REPO/skills/ship-feature/SKILL.md"
+INI="$REPO/skills/init-project/SKILL.md"
+ONB="$REPO/skills/onboard/SKILL.md"
+PASS=0; FAIL=0
+
+tf() { # tf <label> <file> <fixed-string>   (single-line patterns only, LRN-093)
+  if grep -qF -- "$3" "$2" 2>/dev/null; then
+    echo "  PASS $1"; PASS=$((PASS+1))
+  else
+    echo "  FAIL $1 — missing: $3"; FAIL=$((FAIL+1))
+  fi
+}
+
+echo "-- ship-feature (enrich-at-gate) --"
+tf "shf contract step"        "$SHF" "STEP 0e — CONTRACT"
+tf "shf contract-interview"   "$SHF" "lib/contract-interview.md"
+tf "shf enrich at gate"       "$SHF" "ENRICH the STEP 0e contract"
+tf "shf gated marker"         "$SHF" "[gated <date>]"
+tf "shf verify+secure step"   "$SHF" "STEP 5 — VERIFY + SECURE"
+tf "shf uses shared include"  "$SHF" "lib/verify-secure-loop.md"
+tf "shf judges enriched"      "$SHF" "ENRICHED contract"
+tf "shf orthogonal to review" "$SHF" "DISTINCT axis from STEP 6 code review"
+
+echo "-- init-project (contract from BRIEF + adds security) --"
+tf "ini contract from brief"  "$INI" "contract-interview.md"
+tf "ini criteria from V1"     "$INI" "V1 FEATURES (each testable)"
+tf "ini enrich at gate1"      "$INI" "ENRICH the STEP 1 contract"
+tf "ini verify+secure step"   "$INI" "STEP 9 — VERIFY + SECURE"
+tf "ini uses shared include"  "$INI" "lib/verify-secure-loop.md"
+tf "ini adds security gate"   "$INI" "adds the security gate init-project previously lacked"
+
+echo "-- onboard (explicit NO-LOOP audit) --"
+tf "onb no-loop stated"       "$ONB" "n'a PAS de boucle verify"
+tf "onb audit not gate"       "$ONB" "MODE: audit"
+tf "onb scope contract"       "$ONB" "contract de SCOPE"
+tf "onb no symmetry loop"     "$ONB" "Ne PAS ajouter la boucle des flux dev"
+
+echo ""
+echo "loops-heavy structure locks: $PASS pass, $FAIL fail"
+[ "$FAIL" -eq 0 ]

lib/tests/no-vacuous-locks.test.sh

@@ -0,0 +1,55 @@
+#!/usr/bin/env bash
+# ============================================================
+# Deterministic backstop for LRN-093 (vacuous grep locks).
+# grep NEVER interprets a literal \n as a newline: in a -F
+# fixed string it splits the pattern into a per-line OR (matches
+# anything); in -E it is a literal "n". Either way a structure
+# lock carrying \n proves nothing. The LRN advisory alone did not
+# hold (2 recurrences same chantier) -> this mechanical guard.
+#
+# Rule: no backslash-n inside a quoted pattern on a grep / tf /
+# tr_ / tn line, anywhere under lib/tests/*.test.sh. Flip-tested
+# below against a synthetic offender so the guard proves it bites.
+# ============================================================
+set -u
+
+REPO="$(cd "$(dirname "$0")/../.." && pwd)"
+SELF="no-vacuous-locks.test.sh"
+FAIL=0
+
+# One scanner, reused for the real tree and the flip-test fixture.
+# Matches a grep -*F/-*E call OR a tf/tr_/tn helper (at line start or after
+# whitespace) whose quoted pattern contains a literal backslash-n.
+scan() { # scan <dir containing *.test.sh>
+  grep -rnE '(grep +-[A-Za-z]*[EFqe]|(^|[[:space:]])(tf|tr_|tn)[[:space:]]).*"[^"]*\\n' \
+    "$1"/*.test.sh 2>/dev/null | grep -v "$SELF"
+}
+
+echo "-- LRN-093 backstop: scan lib/tests/*.test.sh --"
+HITS="$(scan "$REPO/lib/tests")"
+if [ -n "$HITS" ]; then
+  FAIL=1
+  printf '%s\n' "$HITS" | while IFS= read -r line; do
+    printf '  FAIL vacuous backslash-n lock: %s\n' "$line"
+  done
+else
+  printf '  PASS no vacuous backslash-n locks in lib/tests/*.test.sh\n'
+fi
+
+# Flip-test: the guard MUST catch a known offender (LRN-093 discipline —
+# prove a lock CAN fail before trusting its green).
+echo "-- flip-test: guard bites a synthetic offender --"
+TMP="$(mktemp -d)"
+# shellcheck disable=SC2016  # the single quotes are deliberate: literal backslash-n
+printf '%s\n' 'tf "bad" "$F" "no\nforced loop"' > "$TMP/z.test.sh"
+if [ -n "$(scan "$TMP")" ]; then
+  printf '  PASS guard catches the synthetic offender\n'
+else
+  printf '  FAIL guard blind to a known offender (regex too weak)\n'
+  FAIL=1
+fi
+rm -rf "$TMP"
+
+echo ""
+if [ "$FAIL" -eq 0 ]; then echo "no-vacuous-locks: clean"; else echo "no-vacuous-locks: vacuous locks present"; fi
+[ "$FAIL" -eq 0 ]

skills/init-project/SKILL.md

@@ -48,6 +48,14 @@ ls CLAUDE.md .claude/CLAUDE.md 2>/dev/null | head -1
 
 In both cases: MANDATORY STOP until user answers remaining questions. Produce PROJECT BRIEF.
 
+**Then run `$HOME/.claude/lib/contract-interview.md`** seeded from the BRIEF:
+REQUEST verbatim = the user's project description; ACCEPTANCE CRITERIA = the
+V1 FEATURES (each testable); FILE SCOPE = the planned tree. No new questions
+(the interview already asked). It writes
+`.claude/tasks/contracts/<date>-<slug>-<HHMM>.md`; the DESIGN approved at STEP
+4 ENRICHES it, and STEP 9's verifier judges the MVP against the enriched
+contract.
+
 ## STEP 2 — ANALYZE
 Load `$HOME/.claude/agents/analyzer.md`. Analyze BRIEF: existing code, stack constraints, infra risks, open decisions. Produce ANALYSIS REPORT.
 
@@ -70,6 +78,11 @@ Approve? (yes / request changes)
 ```
 Changes → back to STEP 3. Approved → continue.
 
+**On approval — ENRICH the STEP 1 contract**: append the DESIGN-derived
+acceptance criteria (resolved decisions, interfaces, test strategy) to the
+contract, each tagged `[gated <date>]`. STEP 9's verifier judges against this
+enriched contract.
+
 ## STEP 5 — SCAFFOLD
 Load `$HOME/.claude/agents/scaffolder.md`. Pass: BRIEF + DESIGN + `~/.claude/templates/project-CLAUDE.md` + `~/.claude/CLAUDE.md`.
 Creates: CLAUDE.md, settings, structure, config, empty entry points, .gitignore, .env.example, .claude/tasks/TODO.md, .claude/memory/{decisions,learnings,blockers,journal,evals}.md, .claude/audits/. NO README, NO features.
@@ -175,8 +188,21 @@ If `graphify` CLI is installed AND complexity >= 30%:
 2. Print: `🔗 Full project graph updated at graphify-out/`
 If `graphify` not installed or complexity < 30% → skip silently.
 
-## STEP 9 — ANALYZE
+## STEP 9 — VERIFY + SECURE (fresh gates, bounded loops)
-Load `$HOME/.claude/agents/analyzer.md`. Check: no regressions, no deviations, no stale scaffold, conventions respected.
+Run the two fresh gates per `$HOME/.claude/lib/verify-secure-loop.md` with
+`CONTRACT` = the STEP 1 path (ENRICHED at STEP 4), `DIFF` = the MVP branch
+diff (`develop..HEAD`), `TEST` = the project suite:
+- GATE 1 — a FRESH verifier judges the MVP against the enriched contract (V1
+  features + `[gated]` design criteria). CONFORME → GATE 2. ECARTS → fix,
+  re-verify, max 3 → STOP + human escalation with the CRITERIA table.
+- GATE 2 — a FRESH security-auditor (`MODE: gate`, `SCOPE: develop..HEAD`).
+  PASS → STEP 10. BLOCK → fix, re-verify request THEN re-scan, max 3 →
+  escalate.
+
+This adds the security gate init-project previously lacked (security was only
+deferred to a later /onboard) and turns the informal analyze into a verdict
+against the founding contract. Distinct axis from STEP 10 code review
+([[LRN-095]]) — both run.
 
 ## STEP 10 — CODE REVIEW
 Invoke `superpowers:requesting-code-review`. Fix all CRITICAL before proceeding.

skills/onboard/SKILL.md

@@ -509,6 +509,15 @@ Agent(
 Si semgrep absent → l'agent rend DEGRADED (checklist seule) + recommande
 `make plugin` ; NON bloquant en onboard (audit, pas gate).
 
+**Onboard n'a PAS de boucle verify→dev (`lib/verify-secure-loop.md`) — par
+conception.** onboard produit un RAPPORT d'audit, pas une modification à
+vérifier contre une demande : il n'y a ni contract de conformité, ni diff dev,
+ni verifier, ni max-3. Le contract d'onboard est un contract de SCOPE (ce que
+l'interview STEP 3 + `audit_stack` définissent comme périmètre d'audit), et
+`security-auditor` tourne en `MODE: audit` (report-only), jamais en `MODE:
+gate`. Ne PAS ajouter la boucle des flux dev ici par symétrie — l'audit et le
+flux de dev sont deux formes distinctes ([[BDR-050]] pipeline dev ≠ audit).
+
 #### Dispatch doc-syncer (si `doc` dans audit_stack)
 ```
 Agent(

skills/ship-feature/SKILL.md

@@ -78,7 +78,16 @@ The returned digest (ANALYSIS + RELATED MEMORY) stays in the orchestrator's cont
 is FED to STEP 1 and STEP 2 and reconciled at STEP 3. Degradation: request too vague →
 analyzer flags ambiguous zones, does not block (STEP 1 refines). `.claude/memory/` empty or
 absent → analyzer omits RELATED MEMORY (no-op); the step still returns the code ANALYSIS.
-Additive — distinct from STEP 5 ANALYZE (post-impl regression) and STEP 4b DEBUG.
+Additive — distinct from STEP 5 VERIFY + SECURE (post-impl) and STEP 4b DEBUG.
+
+## STEP 0e — CONTRACT
+
+Run `$HOME/.claude/lib/contract-interview.md`. REQUEST verbatim = the feature
+request as typed; initial ACCEPTANCE CRITERIA from the request; FILE SCOPE
+seeded from 0d's KEY COMPONENTS. It writes
+`.claude/tasks/contracts/<date>-<slug>-<HHMM>.md`; keep the path — the design
+approved at STEP 3 ENRICHES it, and STEP 5's verifier judges the diff against
+the ENRICHED contract. This is the only flow where the contract grows mid-run.
 
 ## STEP 1 — BRAINSTORM
 Invoke `superpowers:brainstorming` — but FEED it the STEP 0d digest as binding context,
@@ -120,6 +129,13 @@ never a guarantee (same discipline as the memory-commit `✅<hash>`: show what's
 assert a check not performed). No RELATED MEMORY from 0d → omit the block.
 Changes → back to STEP 2. Approved → continue.
 
+**On approval — ENRICH the STEP 0e contract.** The design just validated adds
+detail the raw request lacked: append the design-derived acceptance criteria
+to the contract's ACCEPTANCE CRITERIA, each tagged `[gated <date>]` (this is
+the human micro-gate that authorizes contract growth). STEP 5's verifier
+judges the diff against this ENRICHED contract, not the STEP 0e seed — so a
+criterion the design introduced is verified, not lost.
+
 ## STEP 4 — IMPLEMENT
 Start the feature branch off develop, then implement on it:
 ```bash
@@ -157,8 +173,22 @@ OPTIONS :
        Skip them too? (yes / keep and accept partial implementation)"
      If no dependents → skip cleanly and continue.
 
-## STEP 5 — ANALYZE
+## STEP 5 — VERIFY + SECURE (fresh gates, bounded loops)
-Load `$HOME/.claude/agents/analyzer.md`. Check: no regressions, no stale code, no plan deviations.
+Run the two fresh gates per `$HOME/.claude/lib/verify-secure-loop.md` with
+`CONTRACT` = the STEP 0e path (ENRICHED at STEP 3), `DIFF` = the branch diff
+(`develop..HEAD`), `TEST` = the project suite:
+- GATE 1 — a FRESH verifier judges the branch against the ENRICHED contract
+  (all criteria, including the `[gated]` design ones). CONFORME → GATE 2.
+  ECARTS → hand the dev the gap list, fix, re-verify, max 3 → STOP + human
+  escalation with the CRITERIA table.
+- GATE 2 — a FRESH security-auditor (`MODE: gate`, `SCOPE: develop..HEAD`)
+  scans the branch. PASS → STEP 6. BLOCK → fix, re-verify the request THEN
+  re-scan, max 3 → escalate.
+
+This replaces the old informal "analyze for regressions" with a verdict
+against the contract. It is a DISTINCT axis from STEP 6 code review (contract
+conformity + security vs. craft/design) — both run, neither subsumes the
+other ([[LRN-095]]).
 
 ## STEP 6 — CODE REVIEW
 Invoke `superpowers:requesting-code-review`. Fix all CRITICAL before proceeding.