Borrowing it
Nothing to install: this file belongs to apokamo/kaji. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/apokamo/kaji/main/.claude/skills/review-cycle/SKILL.mdgit clone --depth 1 https://github.com/apokamo/kajiWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/apokamo/kaji/review-cycle)<a href="https://agentmods.dev/skills/apokamo/kaji/review-cycle"><img src="https://agentmods.dev/badge/skills/apokamo/kaji/review-cycle/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/apokamo/kaji/review-cycle"><img src="https://agentmods.dev/badge/skills/apokamo/kaji/review-cycle.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00041 | $0.02551 |
| Opus 5 | $0.00020 | $0.01275 |
| Sonnet 5 | $0.00008 | $0.00510 |
| Haiku 4.5 | $0.00004 | $0.00255 |
Grade A, and why
review-cycle scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 194 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Review Cycle
kaji run .kaji/wf/official/dev.yaml <issue_id> --from review-poll --before close を Bash 経由で
起動し、終了後に /issue-close 実行案内を含む verdict を出力する slash command wrapper skill。
--before close で close step の手前で停止するため、本 skill 経由で workflow を回した後は
PASS なら /issue-close <issue_id> を 手動で 実行する運用となる(close まで全自動にしたい
場合は --before close を外し --from review-poll のみで起動する)。
いつ使うか
| タイミング | このスキル |
|---|---|
| PR レビューループを 1 コマンドで自動化したい(close は手動で確認) | ✅ 必須 |
| close まで全自動で進めたい | ❌ 代わりに kaji run .kaji/wf/official/dev.yaml <id> --from review-poll(--before close を付けない) |
provider.type='github' で codex auto-review が走っている環境 |
✅ 前提(review-poll step が auto-review シグナルを監視) |
provider.type='github' 以外(local 等) |
❌ workflow が requires_provider: github で exit 2 |
ワークフロー内の位置: i-pr → [PR 作成] → /review-cycle → (PASS なら手動 /issue-close)
入力
/review-cycle <issue_id>
$ARGUMENTS = <issue_id>(Issue 番号のみ。PR ID は workflow 内のreviewskill が逆引きする)docs/dev/skill-authoring.md§ 手動実行 の規約に従い、slash command の引数は$ARGUMENTSから取得する(shell の positional parameter$1は使わない)- 未指定(
$ARGUMENTSが空)の場合は skill 側でusage: /review-cycle <issue_id>を stderr に出し、kaji runを実行せずに ABORT verdict を stdout に出力する
実行手順
Step 1: 引数の解析
$ARGUMENTS から第 1 トークンを issue_id として取得する。$ARGUMENTS が空、または
第 1 トークンが空文字の場合は kaji run を実行せず、stderr に usage を出して
ABORT verdict を返すこと(後述の「未指定時の ABORT 経路」を参照)。
以下の擬似コードは Claude(agent)が実際に Bash 経由で実行する想定。$ARGUMENTS は
slash command の引数文字列(例: 23)を agent が直接展開する。
set -u # set -e は外す(exit code を明示的に拾う)
# Step 1: 引数チェック
# $ARGUMENTS は slash command 引数文字列。第 1 トークンを issue_id として取り出す。
# `read -r` で空白区切りの先頭を拾うことで、誤って後続引数を含めないようにする。
read -r ISSUE_ID _REST <<<"${ARGUMENTS:-}"
if [ -z "${ISSUE_ID:-}" ]; then
echo "usage: /review-cycle <issue_id>" >&2
cat <<'VERDICT_EOF'
---VERDICT---
status: ABORT
reason: |
Missing required argument: <issue_id>.
evidence: |
$ARGUMENTS was empty or did not contain an issue_id token.
suggestion: |
Re-invoke as /review-cycle <issue_id> (e.g. /review-cycle 23).
---END_VERDICT---
VERDICT_EOF
exit 2
fi
# Step 2: kaji run 起動
# stdout はそのまま流し、stderr のみ tee で capture して
# ^Workflow aborted: シグナル(kaji_harness/cli_main.py:395-401)を後で grep する。
STDERR_LOG=$(mktemp)
trap 'rm -f "$STDERR_LOG"' EXIT
kaji run .kaji/wf/official/dev.yaml "$ISSUE_ID" --from review-poll --before close \
2> >(tee "$STDERR_LOG" >&2)
EXIT=$?
HAS_ABORT_MARKER=0
if grep -q '^Workflow aborted:' "$STDERR_LOG"; then
HAS_ABORT_MARKER=1
fi
# Step 3: 人間向けの先出しメッセージ(verdict ブロック前に出す)
case "$EXIT" in
0)
echo
echo "review-cycle 完了(workflow PASS)。次に /issue-close $ISSUE_ID を実行してください。"
;;
1)
if [ "$HAS_ABORT_MARKER" -eq 1 ]; then
echo "review-cycle が ABORT verdict で終了しました。Issue を確認してください。" >&2
else
echo "review-cycle が exit 1 で終了しましたが、'Workflow aborted:' マーカーが見当たりません。予期しないエラーの可能性があります。stderr を確認してください。" >&2
fi
;;
2)
echo "review-cycle が定義エラー / 設定エラーで終了しました(exit 2)。kaji validate および .kaji/config.toml を確認してください。" >&2
;;
3)
echo "review-cycle が runtime error で終了しました(exit 3)。stderr の traceback を確認してください。" >&2
;;
*)
echo "review-cycle が未知の exit code $EXIT で終了しました。" >&2
;;
esac
# Step 4: verdict ブロック出力(必須 / docs/dev/skill-authoring.md § verdict 出力規約)
if [ "$EXIT" -eq 0 ]; then
cat <<VERDICT_EOF
---VERDICT---
status: PASS
reason: |
review-cycle workflow completed successfully (exit 0).
evidence: |
kaji run .kaji/wf/official/dev.yaml $ISSUE_ID --from review-poll --before close exited with code 0.
suggestion: |
Run /issue-close $ISSUE_ID to merge the PR and clean up.
---END_VERDICT---
VERDICT_EOF
else
# exit != 0 はすべて ABORT に倒す。reason / suggestion で原因と次の手を書き分ける。
case "$EXIT" in
1)
if [ "$HAS_ABORT_MARKER" -eq 1 ]; then
REASON="workflow ABORT verdict (exit 1, 'Workflow aborted:' marker present in stderr)"
SUGG="Inspect the Issue and Issue comments for the failing step's verdict, then decide whether to /pr-fix manually or close the workflow."
else
REASON="exit 1 without 'Workflow aborted:' marker — possibly an unexpected exception in cli_main.py"
SUGG="Check stderr / traceback. Re-execute the failing step manually for diagnosis."
fi
;;
2)
REASON="definition error / config error (exit 2)"
SUGG="Run 'kaji validate .kaji/wf/official/dev.yaml' to surface the YAML or skill error; verify .kaji/config.toml has [provider]."
;;
3)
REASON="runtime error in kaji run (exit 3)"
SUGG="Inspect stderr traceback. The failing CLI dispatch or verdict parse is logged there."
;;
*)
REASON="unknown exit code $EXIT"
SUGG="Inspect kaji run stdout/stderr. This exit code is not defined in docs/dev/workflow-authoring.md § 終了コード."
;;
esac
cat <<VERDICT_EOF
---VERDICT---
status: ABORT
reason: |
$REASON
evidence: |
kaji run .kaji/wf/official/dev.yaml $ISSUE_ID --from review-poll --before close exited with code $EXIT.
Workflow aborted marker in stderr: $HAS_ABORT_MARKER (1 = present, 0 = absent).
suggestion: |
$SUGG
---END_VERDICT---
VERDICT_EOF
fi
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 10d ago First seen · 194 lines · 41 tokens per session scan A 7f8e3c565b69
review-cycle is a skill published in the GitHub repository apokamo/kaji (12 stars, last pushed 2d ago), licensed Apache-2.0. It adds 41 tokens to every session and 2,551 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
gh-find-prs
Survey open Codewhale PRs and triage each for mergeability and disposition against the real landing branch.
merge-queue
Process the Refinery merge queue - collect agent work, detect and resolve conflicts, merge in dependency order, and verify integration.
requesting-code-review
Use when completing tasks, implementing major features, or before merging to verify work meets requirements.
review-open-pull-requests
List and analyze all open pull requests across GitHub repositories. Shows review status, CI/CD check results, age, and reviewers. Use when triaging PRs, checking team velocity, or identifying stale reviews that need attention.
guardian
Gatekeeping Git/PR by classifying change essence and recommending granularity, naming, and strategy. Use when PR preparation or commit strategy is needed.
software-development-norms
Use when writing, reviewing, or refactoring code in any language to apply industrial-grade development norms abstracted from the Alibaba Java Development Handbook. Covers naming, type safety, exceptions, logging, concurrency, databases, project structure, testing, and security with language-agnostic principles and…