kit-skip-review
Records a user's choice to skip the final review gate for the current session.
Install
mkdir -p .claude/skills/kit-skip-review && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/18780" && unzip -o skill.zip -d .claude/skills/kit-skip-review && rm skill.zipInstalls to .claude/skills/kit-skip-review
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Record a USER-approved skip of the kit's final-review Stop gateKey capabilities
- →Record user-approved skips of the final review gate
- →Write a bypass flag for the current session
- →Include the user's exact words in the bypass flag
- →Inform the user that the gate is bypassed for the next turn-end only
How it works
The skill writes a bypass flag to a temporary file, including a timestamp and the user's exact words, which allows the `verify-final-review.sh` Stop hook to be bypassed for one turn.
Inputs & outputs
When to use kit-skip-review
- →Bypass final review gate
- →Record user approval for skip
About this skill
/kit-skip-review — gate skip (user-approved or model-judged)
The Stop hook (verify-final-review.sh) blocks turn-end when business-logic
changes crossed the review threshold. This skill records a decision to waive
that review once. The flag path is in KIT_CONTEXT (session start).
Mode 1 — user mode (the user said to skip)
- Triggered by an explicit user request ("skip review", "這次不用 review").
- Write the flag; first line must start with
user-approved:echo "user-approved date=<YYYY-MM-DD> quote=\"<the user's actual words>\"" > /tmp/claude-skip-review-<session_id>The quote field is the audit trail — the user's real words, never a paraphrase you wish they had said. - This mode can settle ANY batch, including sensitive/protected ones (the gate logs sensitive user-skips to the session skiplog).
Mode 2 — model mode (v4.9: your own judgment, audited)
Use when YOU judge the batch not worth a review: throwaway or experimental code, a one-off analysis script, scratch work that will never be depended on.
- Hard floor (hook re-checks batch content; the flag's self-report counts for nothing): batches touching sensitive paths (auth/payment/migration/security/…), protected-paths, or that are not measurable (no baseline / binary) CANNOT be model-skipped — the gate rejects the flag, audits the attempt, and re-blocks. Those need /kit-review or the user's word.
- Write the flag; both fields are mandatory:
echo "skipped-by=model reason=<one line: why not review-worthy> scope=<N lines/M files>" > /tmp/claude-skip-review-<session_id> - The gate honors it ONLY after appending the reason/scope to the session
skiplog (
/tmp/claude-kit-skiplog-<session_id>.jsonl) — audit-fail-closed. Reasons are sampled against the actual diff; a reason that misdescribes the batch is the fastest way to lose this privilege. - Judgment guide: skip is for "this code does not deserve review", defer is for "review later, when the feature settles". If you hesitate between them, defer (the block message shows the defer one-liner).
After either mode
Tell the user what was skipped and why (one line). The flag is consumed at the next turn-end; later batches are gated again.
When not to use it
- →To self-invoke to get past a block
- →When the user has not explicitly asked to skip the review
- →To end the turn without user approval
Limitations
- →Must be triggered by an explicit user request
- →The bypass flag is consumed after one turn-end
- →The gate only accepts a flag whose first line starts with `user-approved`
How it compares
This skill provides an auditable, user-approved mechanism to bypass a specific review gate for a single turn, unlike simply ignoring the gate or manually editing files.
Compared to similar skills
kit-skip-review side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| kit-skip-review (this skill) | 0 | 2mo | No flags | Beginner |
| conversation-analyzer | 8 | 10mo | Review | Intermediate |
| wrap | 0 | 7mo | Review | Intermediate |
| file-operations | 1 | 10mo | Review | Beginner |
Try saying
Example prompts that trigger this skill in your AI assistant.
You might also like
conversation-analyzer
mhattingpete
Analyzes your Claude Code conversation history to identify patterns, common mistakes, and opportunities for workflow improvement. Use when user wants to understand usage patterns, optimize workflow, identify automation opportunities, or check if they're following best practices.
wrap
team-attention
세션 종료 시 자동 정리를 수행합니다. CLAUDE.md 업데이트 제안, 반복 패턴 자동화 기회 탐지, 배운 것 정리, 미완성 작업 정리. 사용자가 /wrap 또는 세션 마무리 요청 시 사용.
file-operations
mhattingpete
Analyze files and get detailed metadata including size, line counts, modification times, and content statistics. Use when users request file information, statistics, or analysis without modifying files.
triage-issue
mysticaltech
Use when triaging a GitHub issue - analyzes issue, checks for duplicates, categorizes, and drafts response
proof-of-work
MadAppGang
Proof artifact generation patterns for task validation. Covers screenshots, test results, deployments, and confidence scoring.
setup
PackRat-AI
Configure which review agents run for your project. Auto-detects stack and writes compound-engineering.local.md.