thanhnvandClaude Opus 4.8
3adc48ed82
Resolve every root by marker walk-up instead of a fixed depth that only lands on the app via the .specify compat symlink, so the harness runs correctly when invoked by its real packages/casan-harness path — proven by a full gate run via that path: 64/0/0. - 95 scripts/tests: PROJECT_ROOT/ROOT "$SCRIPT_DIR/../.."-style computations -> $CASAN_APP_ROOT. - 6 leaf scripts (infra-lab, context-validate, secrets-scan, path-guard, toolchain-verify, phase2-sourcegen) now source casan-paths + use CASAN_APP_ROOT. - run-casan4: source casan-paths as a package sibling (facade-independent), PROJECT_ROOT=CASAN_APP_ROOT. - 8 Python files: project_root()/REPO_ROOT/bundle_root walk UP for the .specify marker (control-plane-settings, loop_common, model-call, context-compress, test-integrity, bundle-integrity, traceability-matrix; generate-* fixed earlier). - evidence-pack-build.py + traceability-matrix.py: domain refs -> apps/okr/domain (input/, corpus/redteam-vectors.jsonl, traceability-map.json). - ci-harness-gate.sh: export CASAN_TESTS_DIR/CASAN_TEST_MANIFEST/CASAN_BUNDLE_ROOT so the integrity Python resolves via the harness root regardless of invocation path; ROOT=CASAN_APP_ROOT. - Remove the domain compat symlinks from packages/casan-harness/security (redteam-corpus, redteam-vectors, benign-corpus) — packages now holds NO domain data. Both invocation paths pass (compat facade still present): .specify/... and packages/... Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
109 lines
4.5 KiB
Bash
109 lines
4.5 KiB
Bash
#!/usr/bin/env bash
|
|
set -uo pipefail
|
|
|
|
# CASAN Plan-17 Track 4 — Loop Trace / Replay.
|
|
#
|
|
# Proves:
|
|
# * record appends hash-linked Iteration records (append-only),
|
|
# * show renders the loop view (JSON on stdout),
|
|
# * verify-chain confirms an intact chain (exit 0),
|
|
# * editing a recorded record BREAKs the chain (exit 3),
|
|
# * replay re-verifies recorded artifacts and MATCHes when unchanged,
|
|
# * tampering a recorded artifact makes replay DRIFT (verdict mismatch, exit 3),
|
|
# * a corrupt trace file fails closed (verify-chain BREAK),
|
|
# * state redirected -> repo .specify/state stays clean.
|
|
#
|
|
# Deterministic; hermetic; H4 uses the offline pattern layer (no model / network).
|
|
|
|
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
|
|
source "$SCRIPT_DIR/../scripts/bash/casan-paths.sh"
|
|
PROJECT_ROOT="$CASAN_APP_ROOT"
|
|
TRACE="$CASAN_HARNESS_ROOT/scripts/bash/loop-trace.py"
|
|
|
|
WORK="$(mktemp -d)"
|
|
trap 'rm -rf "$WORK"' EXIT
|
|
export CASAN_LOOP_STATE_ROOT="$WORK/state"
|
|
export CASAN_LOOP_POLICY_FILE="$WORK/no-policy.yaml" # strict ceiling; not needed for trace
|
|
|
|
PASS=0; FAIL=0
|
|
pass() { echo "PASS: $1"; PASS=$((PASS + 1)); }
|
|
fail() { echo "FAIL: $1"; FAIL=$((FAIL + 1)); }
|
|
|
|
tr_cmd() {
|
|
# tr_cmd <expect_rc> <desc> -- <args...>; stdout captured in $OUT
|
|
local expect="$1"; shift
|
|
local desc="$1"; shift
|
|
[[ "$1" == "--" ]] && shift
|
|
set +e
|
|
OUT="$(python3 "$TRACE" "$@" 2>/dev/null)"
|
|
local rc=$?
|
|
set -e 2>/dev/null || true
|
|
if [[ "$rc" -eq "$expect" ]]; then pass "$desc (rc=$rc)"; else fail "$desc (rc=$rc, want $expect)"; fi
|
|
}
|
|
|
|
echo "===== Plan-17 T4: Loop Trace / Replay ====="
|
|
|
|
# fixtures: a good artifact that meets criteria (loop-gate PASS)
|
|
printf 'REQ-01 persists OKR to the database. Dashboard renders 3 widgets.\n' > "$WORK/art1.txt"
|
|
printf '{"must_contain":["REQ-01","Dashboard"]}\n' > "$WORK/crit.json"
|
|
|
|
RUN=trace-run-1
|
|
|
|
# 1) record three iterations
|
|
tr_cmd 0 "record step 1" -- record --run-id "$RUN" --step 1 --intent "persist" \
|
|
--action write --tool db --gate-verdict PASS --progress 0.3 --decision CONTINUE \
|
|
--artifact "$WORK/art1.txt" --success-criteria "$WORK/crit.json"
|
|
tr_cmd 0 "record step 2" -- record --run-id "$RUN" --step 2 --intent "render" \
|
|
--action ui --gate-verdict PASS --progress 0.6 --decision CONTINUE
|
|
tr_cmd 0 "record step 3" -- record --run-id "$RUN" --step 3 --intent "done" \
|
|
--gate-verdict PASS --progress 1.0 --decision DONE
|
|
|
|
# 2) show -> JSON loop view with 3 iterations
|
|
tr_cmd 0 "show loop view" -- show --run-id "$RUN"
|
|
grep -q '"count": 3' <<<"$OUT" && pass "show reports 3 iterations" || fail "show wrong count"
|
|
|
|
# 3) verify-chain intact
|
|
tr_cmd 0 "verify-chain intact -> OK" -- verify-chain --run-id "$RUN"
|
|
grep -q '"decision": "OK"' <<<"$OUT" && pass "chain OK decision" || fail "chain not OK"
|
|
|
|
# 4) replay unchanged artifact -> MATCH
|
|
tr_cmd 0 "replay unchanged -> MATCH" -- replay --run-id "$RUN"
|
|
grep -q '"decision": "MATCH"' <<<"$OUT" && pass "replay MATCH" || fail "replay not MATCH"
|
|
|
|
# 5) tamper the recorded artifact -> replay DRIFT (recorded PASS, now FAIL)
|
|
printf 'unrelated content, requirement removed.\n' > "$WORK/art1.txt"
|
|
tr_cmd 3 "replay after artifact tamper -> DRIFT" -- replay --run-id "$RUN"
|
|
grep -q '"decision": "DRIFT"' <<<"$OUT" && pass "replay DRIFT on tamper" || fail "tamper not detected by replay"
|
|
|
|
# 6) tamper a trace record -> verify-chain BREAK
|
|
TRACE_FILE="$CASAN_LOOP_STATE_ROOT/runs/$RUN/trace.jsonl"
|
|
# flip the progress of the 2nd record without recomputing its hash
|
|
python3 - "$TRACE_FILE" <<'PY'
|
|
import json, sys
|
|
p=sys.argv[1]
|
|
lines=[l for l in open(p) if l.strip()]
|
|
e=json.loads(lines[1]); e["iteration"]["progress"]=0.999
|
|
lines[1]=json.dumps(e, sort_keys=True, ensure_ascii=False)+"\n"
|
|
open(p,"w").writelines(lines)
|
|
PY
|
|
tr_cmd 3 "edited record -> chain BREAK" -- verify-chain --run-id "$RUN"
|
|
grep -q '"reason": "chain_broken"' <<<"$OUT" && pass "chain_broken reason" || fail "chain break not reported"
|
|
|
|
# 7) corrupt trace file -> fail-closed BREAK
|
|
RUN2=trace-broken
|
|
mkdir -p "$CASAN_LOOP_STATE_ROOT/runs/$RUN2"
|
|
printf 'not-json{{{\n' > "$CASAN_LOOP_STATE_ROOT/runs/$RUN2/trace.jsonl"
|
|
tr_cmd 3 "corrupt trace -> fail-closed BREAK" -- verify-chain --run-id "$RUN2"
|
|
grep -q '"reason": "fail_closed"' <<<"$OUT" && pass "corrupt trace fails closed" || fail "corrupt trace not fail-closed"
|
|
|
|
# 8) no repo pollution
|
|
if [[ -d "$CASAN_STATE_ROOT/state" ]] && [[ -n "$(ls -A "$CASAN_STATE_ROOT/state" 2>/dev/null)" ]]; then
|
|
fail "repo .specify/state was polluted"
|
|
else
|
|
pass "repo .specify/state stays clean"
|
|
fi
|
|
|
|
echo ""
|
|
echo "===== T4 SUMMARY: PASS=$PASS FAIL=$FAIL ====="
|
|
[[ "$FAIL" -eq 0 ]] || exit 1
|