fix(test): make aggregate runner fail loudly

.machine_readable/scorecards/0-ai-gatekeeper-protocol.scorecard.a2ml

@@ -66,8 +66,8 @@ effects = "/home/user/standards/0-ai-gatekeeper-protocol/CONTRIBUTING.md instruc
 
 [[should]]
 id = "S3"
-text = "The mcp-repo-guardian Deno/TypeScript test suite (manifest_test.js, 36 Deno.test cases covering parsing, security/injection, session and guard logic) SHOULD be executed automatically in CI on every push/PR."
-system = "none (no workflow in .github/workflows/ or .gitlab-ci.yml under mcp-repo-guardian/ invokes `deno test`; deno.json defines a `test` task but nothing calls it in CI)"
+text = "The mcp-repo-guardian Deno test suite (manifest_test.js, 36 Deno.test cases covering parsing, security/injection, session and guard logic) SHOULD be executed automatically in CI on every push/PR."
+system = "Root `just test` invokes `deno test --allow-read test/manifest_test.js` directly; no deno.json task is required. No workflow in .github/workflows/ or .gitlab-ci.yml under mcp-repo-guardian/ invokes that command directly."
 status = "fail"
 effects = "Regressions in the TypeScript/ReScript-compiled manifest/session/guard logic (src/Manifest.mjs, Session.mjs, Guards.mjs) that ships as the actual MCP server enforcement path would not be caught until manually run; Claude/MCP-based consumers of mcp-repo-guardian are the primary enforcement channel per README.adoc, so this is the most consumer-facing untested surface."
 

Justfile

@@ -167,24 +167,14 @@ build:
 
 # Run all sub-project test suites
 test:
-    @echo "=== Standards Monorepo Test Runner ==="
-    @echo ""
-    @echo "[1/5] groove-protocol/reference/ipv6t (Zig — 10 tests)"
-    @cd groove-protocol/reference/ipv6t && zig build test 2>&1 || echo "  SKIP: zig not available"
-    @echo ""
-    @echo "[2/5] 0-ai-gatekeeper-protocol/mcp-repo-guardian (Deno — 36 tests)"
-    @cd 0-ai-gatekeeper-protocol/mcp-repo-guardian && deno task test 2>&1 || echo "  SKIP: deno not available or tests failed"
-    @echo ""
-    @echo "[3/5] axel-protocol (Deno — 14 tests)"
-    @cd axel-protocol && deno task test 2>&1 || echo "  SKIP: deno not available or tests failed"
-    @echo ""
-    @echo "[4/5] a2ml/bindings/rust (Rust — 47 tests)"
-    @cd a2ml/bindings/rust && cargo test 2>&1 || echo "  SKIP: cargo not available or tests failed"
-    @echo ""
-    @echo "[5/5] k9-svc/bindings/rust (Rust — 45 tests)"
-    @cd k9-svc/bindings/rust && cargo test 2>&1 || echo "  SKIP: cargo not available or tests failed"
-    @echo ""
-    @echo "=== Test run complete ==="
+    @bash scripts/run-required-test-suite.sh "MCP repo guardian (Deno — 36 tests)" "0-ai-gatekeeper-protocol/mcp-repo-guardian" deno test --allow-read test/manifest_test.js
+    @bash scripts/run-required-test-suite.sh "Repo guardian offline core (Rust — 29 tests)" "0-ai-gatekeeper-protocol/repo-guardian-fs/tests-offline" cargo test
+    @bash scripts/run-required-test-suite.sh "A2ML Rust binding (47 tests + 3 doctests)" "a2ml/bindings/rust" cargo test
+    @bash scripts/run-required-test-suite.sh "K9 Rust binding (42 tests + 3 doctests)" "k9-svc/bindings/rust" cargo test
+
+# Regression test: test aggregation fails for missing prerequisites and test failures.
+test-runner-test:
+    @bash scripts/tests/run-required-test-suite-test.sh
 
 # Format sub-project code
 fmt:

TEST-NEEDS.adoc

@@ -43,7 +43,7 @@ Zig: grv6; Deno: manifest
 (mcp-repo-guardian, repo-guardian-fs), a2ml, axel-protocol,
 groove-protocol, contractiles, and many more sub-projects
 
-|*Unit tests* |158+ |Real tests across 6 test suites (see breakdown
+|*Runnable automated tests* |154 |Four executable suites (see breakdown
 below)
 
 |*P2P (property) tests* |15+ |proptest in a2ml + k9-svc integration test
@@ -61,13 +61,25 @@ files
 |*Fuzz tests* |0 |Placeholder removed; real fuzz TODO
 |===
 
-=== Test Suite Breakdown (as of 2026-04-04)
+=== Test Suite Breakdown (current runnable suites)
 
-==== groove-protocol/reference/ipv6t — 10 tests (Zig)
+The root `+just test+` aggregate runs the four suites below and fails loudly
+when a required test executable, target directory, or suite fails. It does not
+claim a skipped suite as a pass.
 
-Run: `+zig build test+` from `+groove-protocol/reference/ipv6t/+`
+NOTE: `+build+`, `+fmt+`, and `+lint+` retain their older failure-swallowing
+recipes (and `+build+` still refers to extracted Groove). They are adjacent
+developer-experience debt, deliberately outside this focused test-runner fix.
 
-All 10 tests pass. Cover all 5 spec validation scenarios + 5 property
+==== groove-protocol — extracted from this monorepo
+
+The former `+groove-protocol/reference/ipv6t/+` Zig implementation was
+extracted into the separate `+hyperpolymath/groove-protocol+` repository. It is
+not a standards-local runnable suite.
+
+The following is the historical standards-local description of its 10 tests,
+not a current result for this repository: all 10 covered 5 spec validation
+scenarios + 5 property
 tests: - [x] Positive: correct type hash accepted - [x] Negative: wrong
 type hash rejected before payload parsing - [x] Provenance: 3 chained
 frames produce verifiable hash chain - [x] Fallback: raw bytes without
@@ -84,7 +96,7 @@ hex format: ~57ns/iter (17.5 M/s)
 
 ==== mcp-repo-guardian — 36 tests (Deno/JS)
 
-Run: `+deno task test+` from
+Run: `+deno test --allow-read test/manifest_test.js+` from
 `+0-ai-gatekeeper-protocol/mcp-repo-guardian/+`
 
 All 36 tests pass. Tests cover: - [x] Manifest parsing (hash, canonical
@@ -103,11 +115,14 @@ manifest handled - [x] *E2E dogfood*: parses standards repo’s own
 ~4µs/iter - Full manifest build: ~5.6µs/iter (hash + 5 regex + date) -
 Session lifecycle: ~7.7µs/iter
 
-==== axel-protocol — 14 tests (Deno/TS)
+==== axel-protocol — test infrastructure pending
 
-Run: `+deno task test+` from `+axel-protocol/+`
+The former Deno/TypeScript tests are now AffineScript sources. They are not
+part of `+just test+`: the native AffineScript runner is not yet available and
+the policy-validator fixture still needs an Ajv-compatible test shim. This is
+test-infrastructure debt, not a passing or skipped suite.
 
-All 14 tests pass. Tests cover: - [x] Valid AXEL1 DNS TXT record parsing
+The pending tests cover: - [x] Valid AXEL1 DNS TXT record parsing
 - [x] Extra whitespace handling - [x] Unknown keys ignored - [x] id with
 equals signs (base64) - [x] Reject: missing version, empty payload,
 whitespace-only - [x] Reject: wrong version (AXEL2) - [x] Reject:
@@ -148,7 +163,8 @@ Session: idempotent get_or_create
 
 Run: `+cargo test+` from `+a2ml/bindings/rust/+`
 
-All 47 tests pass (11 inline unit tests + 36 CRG C integration tests).
+All 47 non-documentation tests pass (11 inline unit tests + 36 CRG C
+integration tests), alongside 3 doctests.
 
 CRG C integration tests (`+tests/crg_c_tests.rs+`): - [x] *Smoke*:
 version directive roundtrip, empty document, TrustLevel display (3
@@ -214,7 +230,10 @@ benchmarks (6 bench functions)
 covering all CRG C categories
 * [x] Created `+k9-svc/bindings/rust/benches/k9_bench.rs+` — Criterion
 benchmarks (6 bench functions)
-* [x] All 158+ tests pass across all 6 test suites
+* [x] The four runnable suites (MCP guardian, offline guardian core, A2ML Rust,
+  K9 Rust) pass through the fail-loud `+just test+` aggregate
+* [ ] Restore an executable AXEL AffineScript test harness
+* [ ] Verify the extracted Groove suite in its dedicated repository
 
 === What Was Fixed in Previous Session (2026-04-04, session 1)
 

scripts/run-required-test-suite.sh

@@ -0,0 +1,44 @@
+#!/usr/bin/env bash
+# SPDX-License-Identifier: MPL-2.0
+#
+# Run one required test suite without conflating a missing prerequisite with a
+# passing (or skipped) suite.  The aggregate `just test` recipe is an
+# attestation about these suites, so each prerequisite and test failure must
+# remain observable to its caller.
+
+set -uo pipefail
+
+if [ "$#" -lt 4 ]; then
+  echo "usage: $0 <suite-label> <working-directory> <tool> <tool-arg> [tool-arg ...]" >&2
+  exit 64
+fi
+
+suite_label="$1"
+working_directory="$2"
+tool="$3"
+shift 3
+
+if [ ! -d "$working_directory" ]; then
+  echo "MISSING TARGET: $suite_label: $working_directory is not a directory" >&2
+  exit 66
+fi
+
+if ! command -v "$tool" >/dev/null 2>&1; then
+  echo "UNAVAILABLE: $suite_label: required executable '$tool' is not on PATH" >&2
+  exit 127
+fi
+
+echo "=== $suite_label ==="
+(
+  cd "$working_directory" || exit 66
+  "$tool" "$@"
+)
+status=$?
+
+if [ "$status" -eq 0 ]; then
+  echo "PASS: $suite_label"
+  exit 0
+fi
+
+echo "FAILED (exit $status): $suite_label" >&2
+exit "$status"

scripts/tests/run-required-test-suite-test.sh

@@ -0,0 +1,70 @@
+#!/usr/bin/env bash
+# SPDX-License-Identifier: MPL-2.0
+#
+# Behavioral regression test for the fail-loud aggregate test-suite helper.
+
+set -euo pipefail
+
+ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)"
+RUNNER="$ROOT/scripts/run-required-test-suite.sh"
+TMP="$(mktemp -d)"
+trap 'rm -rf "$TMP"' EXIT
+
+pass=0
+fail=0
+
+ok() {
+  echo "PASS: $1"
+  pass=$((pass + 1))
+}
+
+bad() {
+  echo "FAIL: $1" >&2
+  fail=$((fail + 1))
+}
+
+assert_case() { # label expected_status required_text command...
+  local label="$1" expected_status="$2" required_text="$3"
+  shift 3
+  local output status
+  set +e
+  output="$("$@" 2>&1)"
+  status=$?
+  set -e
+
+  if [ "$status" -eq "$expected_status" ] && printf '%s' "$output" | grep -Fq "$required_text" && ! printf '%s' "$output" | grep -Fq 'SKIP'; then
+    ok "$label"
+  else
+    bad "$label (wanted exit $expected_status and '$required_text'; got exit $status: $output)"
+  fi
+}
+
+mkdir -p "$TMP/suite" "$TMP/bin"
+cat > "$TMP/bin/passing-tool" <<'EOF'
+#!/usr/bin/env bash
+exit 0
+EOF
+cat > "$TMP/bin/failing-tool" <<'EOF'
+#!/usr/bin/env bash
+exit 23
+EOF
+chmod +x "$TMP/bin/passing-tool" "$TMP/bin/failing-tool"
+
+assert_case "available passing suite succeeds" 0 "PASS: passing suite" \
+  env "PATH=$TMP/bin:$PATH" "$RUNNER" "passing suite" "$TMP/suite" passing-tool --test
+assert_case "missing tool fails loudly" 127 "UNAVAILABLE: missing tool suite" \
+  env "PATH=$TMP/bin:$PATH" "$RUNNER" "missing tool suite" "$TMP/suite" absent-tool --test
+assert_case "failing test preserves its exit status" 23 "FAILED (exit 23): failing suite" \
+  env "PATH=$TMP/bin:$PATH" "$RUNNER" "failing suite" "$TMP/suite" failing-tool --test
+assert_case "missing target fails before tool invocation" 66 "MISSING TARGET: missing target suite" \
+  env "PATH=$TMP/bin:$PATH" "$RUNNER" "missing target suite" "$TMP/absent" passing-tool --test
+
+test_recipe="$(sed -n '/^test:/,/^# Regression test/p' "$ROOT/Justfile")"
+if printf '%s\n' "$test_recipe" | grep -Eq '\|\|[[:space:]]*echo.*SKIP|not available or tests failed'; then
+  bad "Justfile does not reintroduce a false-soft test aggregate"
+else
+  ok "Justfile does not reintroduce a false-soft test aggregate"
+fi
+
+echo "run-required-test-suite regression: $pass passed, $fail failed"
+[ "$fail" -eq 0 ]