bigpowers 2.86.6 → 2.86.8

package/scripts/golden-g16-spec-version-gap.sh

@@ -0,0 +1,72 @@
+#!/usr/bin/env bash
+# story: e44s02
+# golden-g16-spec-version-gap.sh — bigpowers' own repo must never show a spec
+# version gap against itself.
+#
+# check-spec-version-gap.sh is a fully-formed SoT-drift detector for consuming
+# projects (installed package.json version vs the spec-format version stamped
+# or fingerprinted in their specs/), the same class of check as G-12, but it
+# had no gate of its own and was reachable from nothing (G-13). Running it
+# against this repo's own specs/ is the natural self-test the tool already
+# supports via --project: bigpowers' own package.json version and its own
+# specs/state.yaml bigpowers_version stamp must always agree.
+#
+# Usage: bash scripts/golden-g16-spec-version-gap.sh [--self-test]
+# Exit 0: no gap (or, under --self-test, a deliberately injected gap was
+#         correctly detected)
+# Exit 1: a real gap exists, or --self-test failed to detect an injected one
+
+set -euo pipefail
+source "$(dirname "${BASH_SOURCE[0]}")/lib/skill-common.sh"
+resolve_repo_root
+cd "$REPO_ROOT"
+
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+NC='\033[0m'
+
+if [[ "${1:-}" == "--self-test" ]]; then
+  echo "=== G-16 self-test: prove the gap detector can fail ==="
+  TMP=$(mktemp -d)
+  trap 'rm -rf "$TMP"' EXIT
+  mkdir -p "$TMP/specs"
+  cp package.json "$TMP/package.json"
+  # A stamped version far behind package.json's must be reported as a gap.
+  # set +e around the call: under set -e, `out=$(cmd)` itself aborts the
+  # script when cmd exits non-zero — exactly the outcome under test here — so
+  # rc=$? would never be reached.
+  printf 'bigpowers_version: 0.0.1\nactive_flow: null\n' > "$TMP/specs/state.yaml"
+  set +e
+  out=$(bash scripts/check-spec-version-gap.sh --project "$TMP" --skip-block --json 2>&1)
+  rc=$?
+  set -e
+  # check-spec-version-gap.sh's stamp path emits json.dumps-formatted JSON
+  # ("gap": true, with a space); its fingerprint path emits hand-concatenated
+  # JSON ("gap":true, no space) — tolerate both rather than assuming one.
+  if [[ $rc -eq 1 ]] && grep -qE '"gap":[[:space:]]*true' <<<"$out"; then
+    echo -e "${GREEN}PASS${NC} injected version gap detected (exit $rc): $out"
+    echo "G-16 self-test: PASS"
+    exit 0
+  fi
+  echo -e "${RED}FAIL${NC} injected gap was not detected (exit $rc): $out"
+  echo "G-16 self-test: FAIL"
+  exit 1
+fi
+
+echo "=== G-16: spec version gap (self-check) ==="
+# Same set -e hazard as the self-test above: a real gap exits non-zero, and
+# without set +e this assignment would abort before the FAIL branch prints.
+set +e
+OUT=$(bash scripts/check-spec-version-gap.sh --project . --skip-block --json 2>&1)
+RC=$?
+set -e
+echo "  $OUT"
+
+if [[ $RC -eq 0 ]]; then
+  echo -e "${GREEN}PASS${NC} no spec version gap"
+  echo "G-16: PASS"
+  exit 0
+fi
+echo -e "${RED}FAIL${NC} bigpowers' own specs/ disagree with its own package.json version (exit $RC)"
+echo "G-16: FAIL"
+exit 1

package/scripts/lib/audit-compliance-report.sh

@@ -5,6 +5,23 @@
 if [ -n "${AUDIT_COMPLIANCE_REPORT_LOADED:-}" ]; then return 0; fi
 AUDIT_COMPLIANCE_REPORT_LOADED=1
 
+# A score sitting exactly on the 94% threshold has zero margin: one more
+# failing scenario (the suite grows over time) blocks all forward work under
+# this repo's own Always-Green doctrine — it already happened once mid-session.
+# pass-warn surfaces a thinning margin before it goes red; it still exits 0,
+# it does not block, it only makes the OKF report and terminal output loud
+# about a score close enough to the threshold to be worth attention.
+audit_gate_status() {
+  local score="$1"
+  if [[ "$score" -lt 94 ]]; then
+    echo "fail"
+  elif [[ "$score" -lt 97 ]]; then
+    echo "pass-warn"
+  else
+    echo "pass"
+  fi
+}
+
 audit_write_okf_report() {
   local OKF_DIR OKF_FILE TIMESTAMP GIT_COMMIT GATE TOTAL_GLOBAL_ALL
 
@@ -20,7 +37,7 @@ audit_write_okf_report() {
   OKF_FILE="$OKF_DIR/audit-$(date +%Y-%m-%d).okf.md"
   TIMESTAMP=$(date -u +"%Y-%m-%dT%H:%M:%SZ")
   GIT_COMMIT=$(git rev-parse HEAD 2>/dev/null || echo "unknown")
-  GATE=$(if [[ $SCORE -ge 94 ]]; then echo "pass"; else echo "fail"; fi)
+  GATE=$(audit_gate_status "$SCORE")
 
   cat > "$OKF_FILE" << OKF_EOF
 ---
@@ -67,10 +84,18 @@ audit_print_summary_and_exit() {
     echo "  SCORE: ${SCORE}% (threshold 94%)"
   fi
 
-  if [[ $SCORE -ge 94 ]]; then
-    echo "  GATE: PASS"
-    exit 0
-  fi
-  echo "  GATE: FAIL (below 94%)"
-  exit 1
+  case "$(audit_gate_status "$SCORE")" in
+    pass)
+      echo "  GATE: PASS"
+      exit 0
+      ;;
+    pass-warn)
+      echo "  GATE: PASS (WARNING: ${SCORE}% is within 3 points of the 94% threshold — margin is thin)"
+      exit 0
+      ;;
+    *)
+      echo "  GATE: FAIL (below 94%)"
+      exit 1
+      ;;
+  esac
 }

package/scripts/lib/audit-compliance-runner.sh

@@ -84,13 +84,26 @@ audit_run_file() {
   TOTAL_PASS=0
   TOTAL_FAIL=0
 
-  while IFS= read -r line; do
+  # `|| [[ -n "$line" ]]` is required: read returns non-zero on a final line
+  # with no trailing newline, so without this a feature file not ending in \n
+  # silently drops its last step — confirmed on karpathy.feature, whose last
+  # line ("And I should push back on unnecessary complexity") never appeared
+  # in [STEP] output at all.
+  while IFS= read -r line || [[ -n "$line" ]]; do
     line=$(echo "$line" | sed 's/^[[:space:]]*//;s/[[:space:]]*$//')
 
     if [[ "$line" =~ ^Feature: ]]; then
       CURRENT_FEATURE="${line#Feature: }"
       echo "FEATURE: $CURRENT_FEATURE"
       echo "## Feature: $CURRENT_FEATURE" >> "$REPORT_FILE"
+    elif [[ "$line" =~ ^Background: ]]; then
+      # A Background's Given/And lines must execute the same as a Scenario's —
+      # without this branch IN_SCENARIO never goes true here, so every step
+      # under Background is silently skipped (not FAILed, not counted at all).
+      # Runs once per file, matching how every other line here executes once
+      # as encountered — not true per-scenario Gherkin re-execution semantics.
+      IN_SCENARIO=true
+      CURRENT_SCENARIO="Background"
     elif [[ "$line" =~ ^Scenario: ]]; then
       CURRENT_SCENARIO="${line#Scenario: }"
       if [[ -n "$SCENARIO_FILTER" && "$CURRENT_SCENARIO" != "$SCENARIO_FILTER" ]]; then

package/scripts/lib/golden-suite-gates.sh

@@ -19,20 +19,36 @@ GOLDEN_GATES=(
   # first so a vacuous gate is caught before its verdict is trusted.
   "g12-status-selftest:bash scripts/golden-g12-status-consistency.sh --self-test:false"
   "g12-status-consistency:bash scripts/golden-g12-status-consistency.sh:false"
-  # Anti-vacuity for the target-contract seam: every contract declared in
-  # targets.yaml must have a real assertion, and an unknown contract must FAIL
-  # rather than SKIP+exit 0 (seven contracts were fail-open before this).
-  # Writing a test must be enough to make it run: G-13 fails if any
-  # scripts/test-*.sh is absent from this list.
-  "g13-test-orphans-selftest:bash scripts/golden-g13-test-orphans.sh --self-test:false"
-  "g13-test-orphans:bash scripts/golden-g13-test-orphans.sh:false"
+  # audit-compliance-runner.sh silently dropped a Background: step and a
+  # feature file's last line when it had no trailing newline — a step exists,
+  # is correct, and never runs is exactly this session's recurring defect
+  # class. Fixture-based regression guard for both, plus a check that a fix
+  # to "make steps execute" doesn't also make a genuinely failing step report
+  # as PASS.
+  "g15-compliance-runner-selftest:bash scripts/golden-g15-compliance-runner-selftest.sh --self-test:false"
+  "g15-compliance-runner:bash scripts/golden-g15-compliance-runner-selftest.sh:false"
+  # Writing a script must be enough to make it run: G-13 fails if any
+  # scripts/*.sh is absent, unreferenced, and untagged bp-manual-utility.
+  # Broadened from scripts/test-*.sh only after a catalog-wide review found 9
+  # more orphans among non-test scripts; renamed to match (golden-g13-test-
+  # orphans.sh -> golden-g13-script-orphans.sh).
+  "g13-script-orphans-selftest:bash scripts/golden-g13-script-orphans.sh --self-test:false"
+  "g13-script-orphans:bash scripts/golden-g13-script-orphans.sh:false"
   # An adapter that defines render_skill() must be dispatched by targets.yaml,
   # or sync renders nothing for it while its install still claims success.
   # docs/WORKFLOWS.md is generated from specs/workflows/*.yaml; a diagram nothing
   # regenerates is a diagram that lies (see specs/archive/assets/README.md).
   "workflow-diagrams-fresh:bash scripts/generate-workflow-diagrams.sh --check:false"
+  # REFERENCE.md nav tables are line-number indexes; hand-maintained they drift
+  # the moment a section grows (all 10 were stale when this gate was added).
+  "reference-nav-fresh:bash scripts/generate-reference-nav.sh --check:false"
   "g14-adapter-dispatch-selftest:bash scripts/golden-g14-adapter-dispatch.sh --self-test:false"
   "g14-adapter-dispatch:bash scripts/golden-g14-adapter-dispatch.sh:false"
+  # check-spec-version-gap.sh is a fully-formed SoT-drift detector for
+  # consuming projects but had no gate of its own (G-13 orphan). Self-checking
+  # bigpowers' own repo against its own package.json is the natural gate.
+  "g16-spec-version-gap-selftest:bash scripts/golden-g16-spec-version-gap.sh --self-test:false"
+  "g16-spec-version-gap:bash scripts/golden-g16-spec-version-gap.sh:false"
   "target-contracts:bash scripts/test-target-contracts.sh:false"
   # Registry-driven replacements for the 25 hand-written per-target scripts.
   "install-hub:bash scripts/test-install-hub.sh:false"

package/scripts/lib/test-assertions.sh

@@ -19,3 +19,7 @@ ta_cleanup() {
 ta_pass() { echo "  PASS $1"; TA_PASS=$((TA_PASS + 1)); }
 
 ta_fail() { echo "  FAIL $1"; TA_FAIL=$((TA_FAIL + 1)); }
+
+# Strip ANSI color codes — used by scripts that grep the colored output of a
+# gate/skill they're testing, where the escape codes would break the match.
+ta_strip_ansi() { sed -E 's/\x1b\[[0-9;]*m//g'; }

package/scripts/snapshot-catalog-baseline.sh

@@ -2,6 +2,10 @@
 # story: e54s01
 # snapshot-catalog-baseline.sh — freeze the live skill catalog into a dated,
 # immutable YAML baseline for e56's reclassification diff.
+# bp-manual-utility: e56 (reclassify-catalog) is still `backlog` in
+# execution-status.yaml — its sole consumer hasn't started, so there is
+# nothing to wire this into yet. Promote to a gate when e56 kicks off; run
+# manually until then: bash scripts/snapshot-catalog-baseline.sh
 set -euo pipefail
 
 SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"

package/scripts/test-adapter-render.sh

@@ -18,17 +18,15 @@ source "$(dirname "${BASH_SOURCE[0]}")/lib/skill-common.sh"
 resolve_repo_root
 cd "$REPO_ROOT"
 
-PASS=0
-FAIL=0
-pass() { echo "PASS: $1"; PASS=$((PASS + 1)); }
-fail() { echo "FAIL: $1" >&2; FAIL=$((FAIL + 1)); }
-
-TMP_ROOT=$(mktemp -d)
-trap 'rm -rf "$TMP_ROOT"' EXIT
-
-
+TA_PASS=0
+TA_FAIL=0
+TA_TMPDIR=""
 
+source "$(dirname "${BASH_SOURCE[0]}")/lib/test-assertions.sh"
+trap ta_cleanup EXIT
 
+TA_TMPDIR=$(mktemp -d)
+TMP_ROOT="$TA_TMPDIR"
 echo "=== test-adapter-render.sh ==="
 
 for adapter in scripts/adapters/*.sh; do
@@ -68,9 +66,9 @@ for adapter in scripts/adapters/*.sh; do
   dest="$TMP_ROOT/$id"
   mkdir -p "$dest"
   if render_into "$dest" "test-skill" && [[ -n "$(find "$dest" -mindepth 1 2>/dev/null)" ]]; then
-    pass "$id: render_skill writes an artifact from a SkillIR"
+    ta_pass "$id: render_skill writes an artifact from a SkillIR"
   else
-    fail "$id: render_skill produced no artifact (${outvars[*]})"
+    ta_fail "$id: render_skill produced no artifact (${outvars[*]})"
   fi
 
   # Path traversal must be refused — this is the security assertion the
@@ -82,14 +80,14 @@ for adapter in scripts/adapters/*.sh; do
   sandbox="$TMP_ROOT/sandbox-$id"
   mkdir -p "$sandbox/out"
   if render_into "$sandbox/out" "../escape"; then
-    fail "$id: accepted a path-traversal skill name"
+    ta_fail "$id: accepted a path-traversal skill name"
   else
-    pass "$id: rejects path traversal in skill name"
+    ta_pass "$id: rejects path traversal in skill name"
   fi
   if [[ -e "$sandbox/escape" ]]; then
-    fail "$id: path traversal escaped into $sandbox"
+    ta_fail "$id: path traversal escaped into $sandbox"
   else
-    pass "$id: traversal wrote nothing outside the destination"
+    ta_pass "$id: traversal wrote nothing outside the destination"
   fi
 
   # Hook bundle + guard behaviour, only where the target ships hooks.
@@ -99,18 +97,18 @@ for adapter in scripts/adapters/*.sh; do
     out="$(echo '{"tool_input":{"command":"git push --force origin main"}}' \
       | bash "$guard" 2>/dev/null || true)"
     grep -q 'block' <<<"$out" \
-      && pass "$id: pre-tool guard blocks a dangerous git command" \
-      || fail "$id: pre-tool guard did not block git push --force"
+      && ta_pass "$id: pre-tool guard blocks a dangerous git command" \
+      || ta_fail "$id: pre-tool guard did not block git push --force"
   fi
 done
 
 # The generic adapter-contract runner stays the source of truth for
 # render_skill/wire_context wiring; assert it is green rather than restating it.
 if bash scripts/test-adapters.sh 2>&1 | grep -q '0 failed'; then
-  pass "test-adapters.sh reports 0 failed across the registry"
+  ta_pass "test-adapters.sh reports 0 failed across the registry"
 else
-  fail "test-adapters.sh reported failures"
+  ta_fail "test-adapters.sh reported failures"
 fi
 
-echo "test-adapter-render: $PASS passed, $FAIL failed"
-[[ "$FAIL" -eq 0 ]] || exit 1
+echo "test-adapter-render: $TA_PASS passed, $TA_FAIL failed"
+[[ "$TA_FAIL" -eq 0 ]] || exit 1

package/scripts/test-completeness-critic-scope.sh

@@ -15,9 +15,12 @@ RED='\033[0;31m'
 GREEN='\033[0;32m'
 NC='\033[0m'
 
-FAILURES=0
-tt_pass() { echo -e "${GREEN}PASS${NC} $1"; }
-tt_fail() { echo -e "${RED}FAIL${NC} $1"; FAILURES=$((FAILURES + 1)); }
+TA_PASS=0
+TA_FAIL=0
+TA_TMPDIR=""
+source "$REPO_ROOT/scripts/lib/test-assertions.sh"
+# NOT `trap ta_cleanup EXIT` — this script's cleanup restores backed-up
+# specs/state.yaml and specs/traceability-matrix.json, not a TA_TMPDIR.
 
 STATE_FILE="specs/state.yaml"
 MATRIX_FILE="specs/traceability-matrix.json"
@@ -26,12 +29,12 @@ MATRIX_BACKUP=$(mktemp)
 cp "$STATE_FILE" "$STATE_BACKUP"
 cp "$MATRIX_FILE" "$MATRIX_BACKUP"
 
-cleanup() {
+completeness_critic_test_cleanup() {
   cp "$STATE_BACKUP" "$STATE_FILE"
   cp "$MATRIX_BACKUP" "$MATRIX_FILE"
   rm -f "$STATE_BACKUP" "$MATRIX_BACKUP"
 }
-trap cleanup EXIT
+trap completeness_critic_test_cleanup EXIT
 
 run_critic() {
   bash scripts/lib/completeness-critic.sh 2>&1 || true
@@ -56,18 +59,18 @@ sed -i.bak '/^active_story:/d' "$STATE_FILE" && rm -f "${STATE_FILE}.bak"
 write_matrix '[]'
 output=$(run_critic)
 if echo "$output" | grep -q "\[WARNING\].*no active_story set"; then
-  tt_pass "unset active_story produces an explicit WARNING"
+  ta_pass "unset active_story produces an explicit WARNING"
 else
-  tt_fail "unset active_story did not produce the expected WARNING"
+  ta_fail "unset active_story did not produce the expected WARNING"
 fi
 
 # --- Scenario: active_story set but absent from the matrix → WARNING ---
 printf 'active_story: e99s99\n' >> "$STATE_FILE"
 output=$(run_critic)
 if echo "$output" | grep -q "\[WARNING\].*e99s99 not found"; then
-  tt_pass "unresolved active_story produces an explicit WARNING"
+  ta_pass "unresolved active_story produces an explicit WARNING"
 else
-  tt_fail "unresolved active_story did not produce the expected WARNING"
+  ta_fail "unresolved active_story did not produce the expected WARNING"
 fi
 sed -i.bak '/^active_story:/d' "$STATE_FILE" && rm -f "${STATE_FILE}.bak"
 
@@ -76,24 +79,20 @@ printf 'active_story: e00s01\n' >> "$STATE_FILE"
 write_matrix '[]'
 output=$(run_critic)
 if echo "$output" | grep -q "\[BLOCKER\] active story e00s01 has zero explicit story-tag links"; then
-  tt_pass "active story with zero tags is correctly flagged as BLOCKER"
+  ta_pass "active story with zero tags is correctly flagged as BLOCKER"
 else
-  tt_fail "active story with zero tags was not flagged as BLOCKER"
+  ta_fail "active story with zero tags was not flagged as BLOCKER"
 fi
 
 # --- Scenario: active story IS tagged → no BLOCKER, even though sibling e00s02 is untagged ---
 write_matrix '[{"method": "explicit_tag", "file": "scripts/fixture.sh", "line": 1, "confidence": 1.0}]'
 output=$(run_critic)
 if echo "$output" | grep -q "\[BLOCKER\] active story e00s01"; then
-  tt_fail "tagged active story incorrectly flagged as BLOCKER"
+  ta_fail "tagged active story incorrectly flagged as BLOCKER"
 else
-  tt_pass "tagged active story is not flagged, and untagged sibling e00s02 does not leak into the check"
+  ta_pass "tagged active story is not flagged, and untagged sibling e00s02 does not leak into the check"
 fi
 
 echo "---"
-if [[ "$FAILURES" -eq 0 ]]; then
-  echo -e "${GREEN}test-completeness-critic-scope: ALL PASS${NC}"
-else
-  echo -e "${RED}test-completeness-critic-scope: $FAILURES failure(s)${NC}"
-fi
-exit "$FAILURES"
+echo "test-completeness-critic-scope: $TA_PASS passed, $TA_FAIL failed"
+exit "$TA_FAIL"

package/scripts/test-generate-epics-wiki-zero-stories.sh

@@ -16,19 +16,22 @@ RED='\033[0;31m'
 GREEN='\033[0;32m'
 NC='\033[0m'
 
-FAILURES=0
-tt_pass() { echo -e "${GREEN}PASS${NC} $1"; }
-tt_fail() { echo -e "${RED}FAIL${NC} $1"; FAILURES=$((FAILURES + 1)); }
+TA_PASS=0
+TA_FAIL=0
+TA_TMPDIR=""
+source "$(dirname "${BASH_SOURCE[0]}")/lib/test-assertions.sh"
+# NOT `trap ta_cleanup EXIT` — this script's cleanup removes a named fixture
+# epic dir and its OKF bundle, not a TA_TMPDIR.
 
 FIXTURE_ID="e00"
 FIXTURE_DIR="specs/epics/e00-zero-story-fixture"
 FIXTURE_BUNDLE="specs/epics-wiki/${FIXTURE_ID}.okf.md"
 
-cleanup() {
+epics_wiki_test_cleanup() {
   rm -rf "$FIXTURE_DIR"
   rm -f "$FIXTURE_BUNDLE"
 }
-trap cleanup EXIT
+trap epics_wiki_test_cleanup EXIT
 
 mkdir -p "$FIXTURE_DIR"
 cat > "${FIXTURE_DIR}/epic.yaml" <<'FIXTUREEOF'
@@ -50,15 +53,11 @@ FIXTUREEOF
 bash scripts/generate-epics-wiki.sh > /dev/null
 
 if [[ ! -f "$FIXTURE_BUNDLE" ]]; then
-  tt_pass "generate-epics-wiki.sh skips a zero-story epic (no bundle emitted)"
+  ta_pass "generate-epics-wiki.sh skips a zero-story epic (no bundle emitted)"
 else
-  tt_fail "generate-epics-wiki.sh emitted a bundle for a zero-story epic: $FIXTURE_BUNDLE"
+  ta_fail "generate-epics-wiki.sh emitted a bundle for a zero-story epic: $FIXTURE_BUNDLE"
 fi
 
 echo "---"
-if [[ "$FAILURES" -eq 0 ]]; then
-  echo -e "${GREEN}test-generate-epics-wiki-zero-stories: ALL PASS${NC}"
-else
-  echo -e "${RED}test-generate-epics-wiki-zero-stories: $FAILURES failure(s)${NC}"
-fi
-exit "$FAILURES"
+echo "test-generate-epics-wiki-zero-stories: $TA_PASS passed, $TA_FAIL failed"
+exit "$TA_FAIL"

package/scripts/test-golden-g11-worktree-venv.sh

@@ -15,10 +15,12 @@ RED='\033[0;31m'
 GREEN='\033[0;32m'
 NC='\033[0m'
 
-FAILURES=0
-tt_pass() { echo -e "${GREEN}PASS${NC} $1"; }
-tt_fail() { echo -e "${RED}FAIL${NC} $1"; FAILURES=$((FAILURES + 1)); }
-strip_ansi() { sed -E 's/\x1b\[[0-9;]*m//g'; }
+TA_PASS=0
+TA_FAIL=0
+TA_TMPDIR=""
+source "$REPO_ROOT/scripts/lib/test-assertions.sh"
+# NOT `trap ta_cleanup EXIT` — this script's cleanup moves a real .venv aside
+# and back, not a TA_TMPDIR.
 
 VENV_MOVED=0
 if [[ -e .venv ]]; then
@@ -26,38 +28,34 @@ if [[ -e .venv ]]; then
   VENV_MOVED=1
 fi
 
-cleanup() {
+golden_g11_test_cleanup() {
   rm -rf .venv
   if [[ "$VENV_MOVED" -eq 1 ]]; then
     mv .venv.test-golden-g11-backup .venv
   fi
 }
-trap cleanup EXIT
+trap golden_g11_test_cleanup EXIT
 
 # --- Scenario: .venv absent (fresh worktree) → gate still PASSes ---
 if [[ -e .venv ]]; then
-  tt_fail "test setup failed to remove .venv before running the gate"
+  ta_fail "test setup failed to remove .venv before running the gate"
 else
-  output=$(bash scripts/golden-g11-gitignore-venv.sh 2>&1 | strip_ansi) && exit_code=0 || exit_code=$?
+  output=$(bash scripts/golden-g11-gitignore-venv.sh 2>&1 | ta_strip_ansi) && exit_code=0 || exit_code=$?
   if [[ "$exit_code" -eq 0 ]] && echo "$output" | grep -q "PASS git check-ignore .venv"; then
-    tt_pass "gate PASSes when .venv does not exist on disk"
+    ta_pass "gate PASSes when .venv does not exist on disk"
   else
-    tt_fail "gate did not PASS with .venv absent (exit=$exit_code)"
+    ta_fail "gate did not PASS with .venv absent (exit=$exit_code)"
   fi
 
   # The gate must clean up the .venv it created for the check — a fresh
   # worktree shouldn't be left with a stray directory after a gate run.
   if [[ ! -e .venv ]]; then
-    tt_pass "gate removes the .venv it created for the check"
+    ta_pass "gate removes the .venv it created for the check"
   else
-    tt_fail "gate left a stray .venv behind after the check"
+    ta_fail "gate left a stray .venv behind after the check"
   fi
 fi
 
 echo "---"
-if [[ "$FAILURES" -eq 0 ]]; then
-  echo -e "${GREEN}test-golden-g11-worktree-venv: ALL PASS${NC}"
-else
-  echo -e "${RED}test-golden-g11-worktree-venv: $FAILURES failure(s)${NC}"
-fi
-exit "$FAILURES"
+echo "test-golden-g11-worktree-venv: $TA_PASS passed, $TA_FAIL failed"
+exit "$TA_FAIL"

package/scripts/test-install-distribute-only.sh

@@ -5,24 +5,25 @@
 set -euo pipefail
 
 REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
-PASS=0
-FAIL=0
+TA_PASS=0
+TA_FAIL=0
+TA_TMPDIR=""
 
-pass() { echo "PASS: $1"; PASS=$((PASS + 1)); }
-fail() { echo "FAIL: $1" >&2; FAIL=$((FAIL + 1)); }
+source "$REPO_ROOT/scripts/lib/test-assertions.sh"
+trap ta_cleanup EXIT
 
 echo "=== test-install-distribute-only.sh ==="
 
 SYNC_SH="$REPO_ROOT/scripts/sync-skills.sh"
 SETUP_JS="$REPO_ROOT/bin/setup.js"
 
-grep -q -- '--distribute-only' "$SYNC_SH" && pass 'sync-skills.sh: --distribute-only flag defined' \
-  || fail 'sync-skills.sh: missing --distribute-only flag'
-grep -q "sync-skills.sh --distribute-only" "$SETUP_JS" && pass 'setup.js: invokes sync-skills.sh with --distribute-only' \
-  || fail 'setup.js: does not pass --distribute-only'
+grep -q -- '--distribute-only' "$SYNC_SH" && ta_pass 'sync-skills.sh: --distribute-only flag defined' \
+  || ta_fail 'sync-skills.sh: missing --distribute-only flag'
+grep -q "sync-skills.sh --distribute-only" "$SETUP_JS" && ta_pass 'setup.js: invokes sync-skills.sh with --distribute-only' \
+  || ta_fail 'setup.js: does not pass --distribute-only'
 grep -q "runInheritedAsync('bash scripts/sync-skills.sh --distribute-only')" "$SETUP_JS" \
-  && pass 'setup.js: sync-skills call is async (spinner can animate)' \
-  || fail 'setup.js: sync-skills call reverted to blocking runInherited (spinner would freeze)'
+  && ta_pass 'setup.js: sync-skills call is async (spinner can animate)' \
+  || ta_fail 'setup.js: sync-skills call reverted to blocking runInherited (spinner would freeze)'
 
 hash_of() { [[ -f "$1" ]] && shasum -a 256 "$1" | awk '{print $1}' || echo "absent"; }
 listing_of() { [[ -d "$1" ]] && (find "$1" -maxdepth 1 -type f | sort) || echo "absent"; }
@@ -34,29 +35,29 @@ EPICS_WIKI_BEFORE=$(listing_of "$REPO_ROOT/specs/epics-wiki")
 
 OUT="$(cd "$REPO_ROOT" && bash scripts/sync-skills.sh --distribute-only 2>&1)"
 echo "$OUT" | grep -q -- '--distribute-only: dev-maintenance regen skipped' \
-  && pass 'sync-skills.sh --distribute-only: confirmation message printed' \
-  || fail 'sync-skills.sh --distribute-only: missing confirmation message'
+  && ta_pass 'sync-skills.sh --distribute-only: confirmation message printed' \
+  || ta_fail 'sync-skills.sh --distribute-only: missing confirmation message'
 echo "$OUT" | grep -q 'Generating OKF wikis' \
-  && fail 'sync-skills.sh --distribute-only: still generated OKF wikis' \
-  || pass 'sync-skills.sh --distribute-only: no OKF wiki generation'
+  && ta_fail 'sync-skills.sh --distribute-only: still generated OKF wikis' \
+  || ta_pass 'sync-skills.sh --distribute-only: no OKF wiki generation'
 echo "$OUT" | grep -q 'regenerate-lockfile' \
-  && fail 'sync-skills.sh --distribute-only: still regenerated lockfile' \
-  || pass 'sync-skills.sh --distribute-only: no lockfile regen'
+  && ta_fail 'sync-skills.sh --distribute-only: still regenerated lockfile' \
+  || ta_pass 'sync-skills.sh --distribute-only: no lockfile regen'
 
 [[ "$(hash_of "$REPO_ROOT/skills-lock.json")" == "$LOCK_BEFORE" ]] \
-  && pass 'skills-lock.json untouched' || fail 'skills-lock.json was modified'
+  && ta_pass 'skills-lock.json untouched' || ta_fail 'skills-lock.json was modified'
 [[ "$(hash_of "$REPO_ROOT/SKILL-INDEX.md")" == "$INDEX_BEFORE" ]] \
-  && pass 'SKILL-INDEX.md untouched' || fail 'SKILL-INDEX.md was modified'
+  && ta_pass 'SKILL-INDEX.md untouched' || ta_fail 'SKILL-INDEX.md was modified'
 [[ "$(hash_of "$REPO_ROOT/specs/SKILL-SEARCH-INDEX_LATEST.md")" == "$SEARCH_INDEX_BEFORE" ]] \
-  && pass 'SKILL-SEARCH-INDEX_LATEST.md untouched' || fail 'SKILL-SEARCH-INDEX_LATEST.md was modified'
+  && ta_pass 'SKILL-SEARCH-INDEX_LATEST.md untouched' || ta_fail 'SKILL-SEARCH-INDEX_LATEST.md was modified'
 [[ "$(listing_of "$REPO_ROOT/specs/epics-wiki")" == "$EPICS_WIKI_BEFORE" ]] \
-  && pass 'specs/epics-wiki/ untouched' || fail 'specs/epics-wiki/ was modified'
+  && ta_pass 'specs/epics-wiki/ untouched' || ta_fail 'specs/epics-wiki/ was modified'
 
-[[ -f "$REPO_ROOT/.cursor/rules/fix-bug.mdc" ]] && pass 'skills still distributed (.cursor/rules populated)' \
-  || fail 'skill distribution did not run'
+[[ -f "$REPO_ROOT/.cursor/rules/fix-bug.mdc" ]] && ta_pass 'skills still distributed (.cursor/rules populated)' \
+  || ta_fail 'skill distribution did not run'
 
-bash -n "$SYNC_SH" && pass 'bash -n sync-skills.sh' || fail 'bash -n sync-skills.sh'
-node --check "$SETUP_JS" && pass 'node --check setup.js' || fail 'node --check setup.js'
+bash -n "$SYNC_SH" && ta_pass 'bash -n sync-skills.sh' || ta_fail 'bash -n sync-skills.sh'
+node --check "$SETUP_JS" && ta_pass 'node --check setup.js' || ta_fail 'node --check setup.js'
 
-echo "test-install-distribute-only: $PASS passed, $FAIL failed"
-[[ "$FAIL" -eq 0 ]] || exit 1
+echo "test-install-distribute-only: $TA_PASS passed, $TA_FAIL failed"
+[[ "$TA_FAIL" -eq 0 ]] || exit 1