Guard commit messages against session metadata

.github/workflows/repository-policy.yml

@@ -16,10 +16,15 @@ jobs:
     steps:
       - name: Check out repository
         uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
 
       - name: Check tracked text file integrity
         run: python3 scripts/tests/check_text_integrity.py
 
+      - name: Check commit messages
+        run: python3 scripts/tests/check_commit_messages.py
+
       - name: Check PR base is current
         if: github.event_name == 'pull_request'
         run: python3 scripts/tests/check_pr_base_current.py

docs/ci.md

@@ -16,8 +16,8 @@ Benchkit uses GitHub as the public development repository and GitLab CI for benc
 | `Sync protected branches to GitLab` | Pushes to `develop` or `main` / `develop`または`main`へのpush | Mirrors protected branches to GitLab without starting GitLab CI / GitLab CIを発火させずに保護ブランチをGitLabへ同期する |
 | `Guard main PR source` | Pull requests to `main` / `main`宛PR | Allows only upstream `develop` to target `main` / upstreamの`develop`から`main`へのPRだけを許可する |
 | `Result Server Tests` | All pull requests and pushes to `develop` or `main` / 全pull requestと`develop`または`main`へのpush | Runs site config preflight, result server tests, and profiler/profile-data/estimation shell tests / site config preflight、result server test、profiler/profile-data/estimation shell testを実行する |
-| `Shellcheck` | Shell script changes under `scripts/`, `programs/`, or `benchpark-bridge/scripts/` / `scripts/`、`programs/`、または`benchpark-bridge/scripts/`配下のshell script変更 | Gates `bash -n` syntax failures and shellcheck error-level issues / `bash -n`構文エラーとshellcheckのerror級をgateする |
-| `Repository Policy` | All pull requests and pushes to `develop` or `main` / 全pull requestと`develop`または`main`へのpush | Validates tracked text files as UTF-8 without replacement characters / 管理下text fileがUTF-8かつreplacement characterなしであることを検証する |
+| `Shellcheck` | Shell script changes under `scripts/`, `programs/`, or `benchpark-bridge/scripts/` / `scripts/`、`programs/`、または`benchpark-bridge/scripts/`配下のshell script変更 | Gates `bash -n` syntax failures and shellcheck error-level issues; pull requests also run warning-level shellcheck for changed app scripts / `bash -n`構文エラーとshellcheckのerror級をgateする。pull requestでは変更されたapp scriptをwarning級でも確認する |
+| `Repository Policy` | All pull requests and pushes to `develop` or `main` / 全pull requestと`develop`または`main`へのpush | Validates tracked text files, commit-message metadata, and current PR base / 管理下text file、commit message metadata、PR baseが最新であることを検証する |
 
 ## GitLab Secrets / GitLab secret
 
@@ -284,9 +284,9 @@ cache miss の場合は通常の `programs/<code>/build.sh` 経路に戻り、
 
 ## Lightweight Repository Policy / 軽量repository policy
 
-`Repository Policy` runs on every pull request and on pushes to `develop` or `main`. It runs `scripts/tests/check_text_integrity.py` to ensure tracked text-like files decode as UTF-8 and do not contain the U+FFFD replacement character.
+`Repository Policy` runs on every pull request and on pushes to `develop` or `main`. It runs `scripts/tests/check_text_integrity.py` to ensure tracked text-like files decode as UTF-8 and do not contain the U+FFFD replacement character. It also runs `scripts/tests/check_commit_messages.py` to reject external AI tool session URLs and private tooling session identifiers in newly introduced commit messages. On pull requests, `scripts/tests/check_pr_base_current.py` confirms that the CI run used the current base branch SHA.
 
-`Repository Policy` は全pull requestと `develop` / `main` へのpushで動きます。`scripts/tests/check_text_integrity.py` を実行し、管理下のtext系fileがUTF-8としてdecodeでき、U+FFFD replacement characterを含まないことを確認します。
+`Repository Policy` は全pull requestと `develop` / `main` へのpushで動きます。`scripts/tests/check_text_integrity.py` を実行し、管理下のtext系fileがUTF-8としてdecodeでき、U+FFFD replacement characterを含まないことを確認します。さらに `scripts/tests/check_commit_messages.py` で、新規に入るcommit messageに外部AI toolのsession URLやprivate tooling session識別子が含まれないことを確認します。pull requestでは `scripts/tests/check_pr_base_current.py` により、CI runが最新のbase branch SHAを使っていることも確認します。
 
 The periodic review should therefore focus on semantic checks that are hard to encode in CI: stale documentation, implementation/design mismatches, workflow review, unused-code candidates, ownership-boundary drift, and research-snapshot consistency. It should not need to repeat the lightweight executable checks unless a CI result itself looks suspicious.
 
@@ -300,7 +300,7 @@ The periodic review should therefore focus on semantic checks that are hard to e
 | `result_server/**/*` / `result_server/**/*` | `Result Server Tests` | Skipped by `.gitlab-ci.yml` rules / `.gitlab-ci.yml` rulesでskip | Portal regressions should be caught by lightweight Python tests / portal回帰はlightweight Python testで捕捉する |
 | Public site config or portal metadata `config/system.csv`, `config/queue.csv`, `config/system_info.csv` / 公開site configまたはportal表示メタデータ`config/system.csv`、`config/queue.csv`、`config/system_info.csv` | `Result Server Tests`, including site config preflight / site config preflightを含む`Result Server Tests` | `config/system.csv` and `config/queue.csv` run by `.gitlab-ci.yml`; `config/system_info.csv` is skipped / `config/system.csv`と`config/queue.csv`は`.gitlab-ci.yml`で実行、`config/system_info.csv`はskip | Public systems listed in `system_info.csv` must also exist in `system.csv` and reference a queue defined in `queue.csv` / `system_info.csv`に載せる公開systemは`system.csv`にも存在し、`queue.csv`定義済みqueueを参照する必要がある |
 | Portal upload or profile-data helper `scripts/bk_functions.sh`, `scripts/result.sh`, `scripts/result_server/**` / portal uploadまたはprofile-data helper `scripts/bk_functions.sh`、`scripts/result.sh`、`scripts/result_server/**` | `Result Server Tests`; `Shellcheck` for `.sh` changes / `Result Server Tests`; `.sh`変更は`Shellcheck` | GitHub pull requests do not start GitLab by default; if a direct/manual GitLab pipeline is started, `scripts/**/*` is treated as benchmark-affecting and runs / GitHub pull requestでは既定でGitLabは起動しない。直接/手動GitLab pipelineを起動した場合、`scripts/**/*` はbenchmark影響ありとして実行される | These helpers shape result JSON / upload behavior. Use lightweight tests first, then start `GitLab Manual CI` when benchmark-side behavior needs validation / これらのhelperはResult JSONやupload挙動へ影響する。まずlightweight testで確認し、benchmark側挙動の検証が必要な場合は`GitLab Manual CI`を起動する |
-| Benchmark app entrypoints `programs/**/build.sh`, `programs/**/estimate.sh`, `programs/**/run.sh`, or app matrix `programs/**/list.csv` / benchmark app entrypoint `programs/**/build.sh`、`programs/**/estimate.sh`、`programs/**/run.sh`、またはapp matrix `programs/**/list.csv` | `Result Server Tests` for portal app-support visibility and diagnostics; `Shellcheck` for `.sh` changes / portal app-support表示とdiagnostics向けに`Result Server Tests`; `.sh`変更は`Shellcheck` | Run through `GitLab Manual CI` when maintainer starts it / maintainerが`GitLab Manual CI`を起動した場合に実行 | App support and diagnostics read list/build/estimate/run for `/results/usage`; these checks provide visibility, not a readiness gate / app supportとdiagnosticsは`/results/usage`用にlist/build/estimate/runを読む。これはvisibilityでありreadiness gateではない |
+| Benchmark app entrypoints `programs/**/build.sh`, `programs/**/estimate.sh`, `programs/**/run.sh`, or app matrix `programs/**/list.csv` / benchmark app entrypoint `programs/**/build.sh`、`programs/**/estimate.sh`、`programs/**/run.sh`、またはapp matrix `programs/**/list.csv` | `Result Server Tests` for portal app-support visibility and diagnostics; `Shellcheck` for `.sh` changes, with PR app-script changes checked at warning level / portal app-support表示とdiagnostics向けに`Result Server Tests`; `.sh`変更は`Shellcheck`、PRのapp script変更はwarning級でも確認 | Run through `GitLab Manual CI` when maintainer starts it / maintainerが`GitLab Manual CI`を起動した場合に実行 | App support and diagnostics read list/build/estimate/run for `/results/usage`; these checks provide visibility, not a readiness gate / app supportとdiagnosticsは`/results/usage`用にlist/build/estimate/runを読む。これはvisibilityでありreadiness gateではない |
 | Other benchmark app files, legacy Benchpark bridge, or shared scripts / その他のbenchmark appファイル、legacy Benchpark bridge、または共通script | `Shellcheck` for `.sh` changes when covered; otherwise normal GitHub review checks / 対象`.sh`変更は`Shellcheck`; それ以外は通常のGitHub review check | Run through `GitLab Manual CI` when maintainer starts it / maintainerが`GitLab Manual CI`を起動した場合に実行 | Use `code` and `system` for Benchkit validation. Use `benchpark` or `park_only` only for the legacy bridge path when needed / Benchkit検証には`code`と`system`を使う。legacy bridge pathが必要な場合だけ`benchpark`または`park_only`を使う |
 | GitHub workflow/action `.github/**/*` / GitHub workflow/action `.github/**/*` | `Repository Policy`; workflow-specific checks when applicable / `Repository Policy`; 必要に応じてworkflow固有check | Skipped by `.gitlab-ci.yml` rules / `.gitlab-ci.yml` rulesでskip | GitHub workflow/action changes affect API-calling or sync control logic. Review workflow semantics carefully; protected-branch sync pushes them to GitLab with `ci.skip` / GitHub workflow/action変更はAPI呼び出しやsync制御に影響する。workflowの意味を慎重にreviewする。protected-branch syncでは`ci.skip`付きでGitLabへpushされる |
 | `.gitlab-ci.yml` / `.gitlab-ci.yml` | Normal GitHub review checks only / 通常のGitHub review checkのみ | Run through `GitLab Manual CI` when a maintainer needs to validate GitLab pipeline behavior / GitLab pipeline挙動の検証が必要な場合にmaintainerが`GitLab Manual CI`で実行 | This file defines GitLab benchmark pipeline behavior / このファイルはGitLab benchmark pipeline挙動を定義する |
@@ -318,7 +318,7 @@ Use these examples when deciding whether to split a pull request or start GitLab
 | `config/system_info.csv` only / `config/system_info.csv`のみ | `Result Server Tests` should verify public site config consistency / 公開site config整合性を`Result Server Tests`で確認 | No benchmark run because this file is portal display metadata / portal表示metadataなのでbenchmark不要 |
 | `config/system.csv` or `config/queue.csv` for a public system / 公開system向けの`config/system.csv`または`config/queue.csv` | `Result Server Tests` should run the site config preflight / `Result Server Tests`でsite config preflightを実行 | Start `GitLab Manual CI` too when benchmark execution behavior needs validation / benchmark実行挙動の検証が必要なら`GitLab Manual CI`も起動 |
 | `scripts/bk_functions.sh`, `scripts/result.sh`, or `scripts/result_server/**` only / `scripts/bk_functions.sh`、`scripts/result.sh`、または`scripts/result_server/**`のみ | `Result Server Tests` should run; `Shellcheck` should run for `.sh` changes / `Result Server Tests`が動く; `.sh`変更では`Shellcheck`が動く | Protected-branch sync uses `ci.skip`; direct/manual GitLab pipelines run because `.gitlab-ci.yml` treats `scripts/**/*` as benchmark-affecting / protected branch syncは`ci.skip`を使う。直接/手動GitLab pipelineでは`.gitlab-ci.yml`が`scripts/**/*`をbenchmark影響ありとして扱うため実行される |
-| `programs/**/build.sh`, `programs/**/estimate.sh`, `programs/**/run.sh`, or `programs/**/list.csv` / `programs/**/build.sh`、`programs/**/estimate.sh`、`programs/**/run.sh`、または`programs/**/list.csv` | `Result Server Tests` should run for app-support/diagnostics visibility; `Shellcheck` should run for `.sh` changes / app-support/diagnostics visibility向けに`Result Server Tests`が動く; `.sh`変更では`Shellcheck`が動く | Start `GitLab Manual CI` when benchmark validation is needed, preferably with explicit `code` and `system` filters / benchmark検証が必要なら`code`と`system`を明示して`GitLab Manual CI`を起動 |
+| `programs/**/build.sh`, `programs/**/estimate.sh`, `programs/**/run.sh`, or `programs/**/list.csv` / `programs/**/build.sh`、`programs/**/estimate.sh`、`programs/**/run.sh`、または`programs/**/list.csv` | `Result Server Tests` should run for app-support/diagnostics visibility; `Shellcheck` should run for `.sh` changes, including warning-level checks for changed app scripts on PRs / app-support/diagnostics visibility向けに`Result Server Tests`が動く; `.sh`変更では`Shellcheck`が動き、PRのapp script変更はwarning級でも確認する | Start `GitLab Manual CI` when benchmark validation is needed, preferably with explicit `code` and `system` filters / benchmark検証が必要なら`code`と`system`を明示して`GitLab Manual CI`を起動 |
 | `scripts/job_functions.sh` or `scripts/test_submit.sh` / `scripts/job_functions.sh`または`scripts/test_submit.sh` | `Shellcheck` should run for `.sh` changes / `.sh`変更では`Shellcheck`が動く | Start `GitLab Manual CI` when scheduler generation or submission behavior needs validation / scheduler生成や投入挙動の検証が必要なら`GitLab Manual CI`を起動 |
 | `benchpark-bridge/scripts/**/*.sh` / `benchpark-bridge/scripts/**/*.sh` | `Shellcheck` should run / `Shellcheck`が動く | Start `GitLab Manual CI` with `benchpark=true` or `park_only=true` only when the legacy bridge behavior needs validation / legacy bridge挙動の検証が必要な場合だけ`benchpark=true`または`park_only=true`で`GitLab Manual CI`を起動 |
 | `.github/workflows/sync-to-gitlab.yml` or `.github/actions/prepare-gitlab-repo/action.yml` / `.github/workflows/sync-to-gitlab.yml`または`.github/actions/prepare-gitlab-repo/action.yml` | `Repository Policy` and GitHub Actions-side validation / `Repository Policy`とGitHub Actions側の確認 | Skipped by `.gitlab-ci.yml` rules when changed alone; protected-branch sync pushes it with `ci.skip` / 単独変更なら`.gitlab-ci.yml` rulesでskip。protected-branch syncでは`ci.skip`付きでpushされる |
@@ -342,3 +342,11 @@ To avoid unnecessary heavy CI runs:
 pull request、issue、commit message、CI設定には、secretやprivate credentialを書かないでください。
 
 Do not include secrets or private credentials in pull requests, issues, commit messages, or CI configuration.
+
+外部AI toolやprivate toolingのsession URLを、commit message、pull request、issue、CI logに含めないでください。commit attributionが必要な場合は、公開してよい一般的な署名や `Co-authored-by` trailer に留めます。
+
+Do not include external AI tool or private tooling session URLs in commit messages, pull requests, issues, or CI logs. If attribution is needed, use public-safe signing or a `Co-authored-by` trailer instead.
+
+このcheckは新規に入るcommitを対象とし、既存履歴を遡及的に書き換えるためのpolicyではありません。
+
+This check applies to newly introduced commits and is not a retroactive history-rewrite policy.

scripts/tests/check_commit_messages.py

@@ -0,0 +1,107 @@
+#!/usr/bin/env python3
+"""Reject commit messages that expose external AI tool session metadata."""
+
+from __future__ import annotations
+
+import json
+import os
+import re
+import subprocess
+import sys
+from pathlib import Path
+
+
+ZERO_SHA = "0" * 40
+DISALLOWED_PATTERNS = (
+    (re.compile(r"(?im)^Claude-Session\s*:"), "Claude-Session trailer"),
+    (re.compile(r"(?i)\b(?:https?://)?claude\.ai/code/session[^\s]*"), "claude.ai session URL"),
+)
+
+
+def run_git(*args: str) -> str:
+    return subprocess.check_output(["git", *args], text=True).strip()
+
+
+def git_fetch(*refspecs: str) -> None:
+    subprocess.run(["git", "fetch", "--no-tags", "origin", *refspecs], check=True)
+
+
+def event_payload() -> dict:
+    event_path = os.environ.get("GITHUB_EVENT_PATH")
+    if not event_path:
+        return {}
+    with Path(event_path).open(encoding="utf-8") as handle:
+        return json.load(handle)
+
+
+def changed_commits() -> list[str]:
+    event_name = os.environ.get("GITHUB_EVENT_NAME")
+    event = event_payload()
+
+    if event_name == "pull_request":
+        pull_request = event.get("pull_request") or {}
+        number = pull_request.get("number")
+        base = pull_request.get("base") or {}
+        base_ref = base.get("ref")
+        base_sha = base.get("sha")
+        if not number or not base_ref or not base_sha:
+            raise RuntimeError("pull_request.number, base.ref, or base.sha is missing")
+
+        head_ref = f"refs/remotes/origin/pr/{number}/head"
+        git_fetch(
+            f"+refs/heads/{base_ref}:refs/remotes/origin/{base_ref}",
+            f"+refs/pull/{number}/head:{head_ref}",
+        )
+        revspec = f"{base_sha}..{head_ref}"
+        return run_git("rev-list", "--reverse", revspec).splitlines()
+
+    if event_name == "push":
+        before = event.get("before")
+        after = event.get("after")
+        if not after or after == ZERO_SHA:
+            return []
+        if not before or before == ZERO_SHA:
+            return [after]
+        return run_git("rev-list", "--reverse", f"{before}..{after}").splitlines()
+
+    return [run_git("rev-parse", "HEAD")]
+
+
+def disallowed_metadata(message: str) -> list[str]:
+    labels: list[str] = []
+    for pattern, label in DISALLOWED_PATTERNS:
+        if pattern.search(message):
+            labels.append(label)
+    return labels
+
+
+def main() -> int:
+    try:
+        commits = changed_commits()
+    except (OSError, subprocess.CalledProcessError, RuntimeError) as exc:
+        print(f"Could not determine commits to check: {exc}", file=sys.stderr)
+        return 1
+
+    violations: list[tuple[str, str]] = []
+    for commit in commits:
+        message = run_git("log", "-1", "--format=%B", commit)
+        for label in disallowed_metadata(message):
+            violations.append((commit, label))
+
+    if violations:
+        print("Disallowed commit metadata detected:", file=sys.stderr)
+        for commit, label in violations:
+            print(f"  - {commit[:12]}: {label}", file=sys.stderr)
+        print(
+            "Remove external AI tool session URLs or private tooling session "
+            "identifiers from commit messages before merging.",
+            file=sys.stderr,
+        )
+        return 1
+
+    print(f"Checked {len(commits)} commit message(s)")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())