菜单
The cipherlib verification gate — run after completing any change and before declaring it done. Ordered, fail-fast, CI-parity checks with a pass/fail report. Also covers when to add node-platform and integration runs.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Run the cipherlib release ritual — local pre-flight mirroring release-test.yml, CHANGELOG entry, single version-bump commit, matching vX.Y.Z tag, tag push, and CI watch through pub.dev publish. Use ONLY when the maintainer explicitly asks to release, publish, or bump the version.
Scaffold a new cipher algorithm, AES mode, or AEAD variant end-to-end in cipherlib house style — implementation, barrel wiring, canonical test file with KAT placeholders, integration parity, benchmark, README row, and CHANGELOG entry. Use whenever adding any new algorithm to this package.
Run cipherlib benchmarks and regenerate BENCHMARK.md plus its README embed safely, with the reference-machine guard. Use when measuring a performance change or when benchmark tables need refreshing.
| name | verify |
| description | The cipherlib verification gate — run after completing any change and before declaring it done. Ordered, fail-fast, CI-parity checks with a pass/fail report. Also covers when to add node-platform and integration runs. |
Run the steps in this order (cheap and common failures first). Stop at the first failure, fix, and restart from that step. Report the full matrix at the end. All commands from the repo root.
# 1. Formatting (CI job: lint)
dart format --output=none --set-exit-if-changed .
# 2. Analyzer at CI strictness — plain `dart analyze` is NOT enough
dart analyze --fatal-infos
# 3. Full VM test suite — never just the touched file; cross-cutting tests
# (cipher_test, compare_test, correctness_test, nonce_test, padding_test)
# exercise shared code paths
dart test -p vm
# 4. Integration parity suite (separate pub package)
cd test_integration && dart pub get && dart run main.dart && cd ..
# 5. Smoke-run the main example
dart run example/cipherlib_example.dart
>>>/bit
tricks, dart:io, or a reported JS bug): also run dart test -p node.
JS ints are not VM ints; shifts and masks near 32/53 bits behave
differently — this is exactly where node catches what vm cannot.benchmark skill
(measurement-only unless on the reference M3 Pro).dart pub global run pana --exit-code-threshold 0
to catch scoring regressions before they surface at release time.If the change alters runtime behavior, exercise it directly: write a 5-line scratch program in the session scratchpad (not in the repo) that drives the changed path and print-compare against a spec vector or a prior-version output. For crypto code, a green round-trip is NOT verification — only a known-answer vector proves the keystream (AGENTS.md §0).
End with an explicit matrix, e.g.:
format PASS
analyze PASS (--fatal-infos)
test -p vm PASS (N tests)
integration PASS
example PASS
node SKIPPED (change not web-relevant)
KAT coverage RFC-8439 §2.3.2 vectors cover the changed path
Never report "done" with a failing or skipped-without-reason row. If a
failure is pre-existing on main (verify by git stash + rerun), say so
explicitly and leave it unfixed (AGENTS.md §6, Flag-2).