메뉴
Stop-hook interception for non-tool responses that claim completion without verification evidence.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Evidence-gated completion check. Use before committing, creating PRs, or claiming work is done. Pairs with verification-spec at task start.
Systematic knowledge extraction from open-source projects. Use when: user shares a repo/link to study, says 偷师/steal/学习/研究, or wants to analyze another project's patterns for adoption.
Launch a real interactive claude CLI session in a target directory (typically a worktree) with a pre-seeded prompt. Use when the current session cannot absorb the task context — e.g., N parallel worktree jobs that would blow the context window, or when you need the user to watch progress in a visible terminal. Each spawned session runs independently in its own Windows Terminal tab.
FileNotFound recovery protocol. Before reporting a missing-file error, suggest near-misses by basename similarity (difflib threshold 0.4) so typos and stale paths fail loudly with actionable hints.
Paste-by-reference guard for Edit calls. When new_string would echo >30 lines of existing file content, use {{file:path:start:end}} reference instead and expand before submission.
File-IO protocol for parent ↔ subagent communication. Parent writes input.txt + context.json, subagent appends to output.txt with [ROUND END] sentinels. Used with Monitor tool for live observation.
| name | no-tool-interception |
| description | Stop-hook interception for non-tool responses that claim completion without verification evidence. |
You are a stop-gate enforcer. Your job is to intercept agent responses that claim task completion without providing verification evidence. Three specific response shapes trigger a block.
The no-tool-gate.sh Stop hook reads last_assistant_message from stdin JSON and checks for three intercept patterns:
Triggers when: Response contains a completion signal matching (任务完成|task complete|完成了|搞定|all done|done\.) (case-insensitive) AND lacks [VERIFY] or VERDICT: token.
Block message: [no-tool-gate] 检测到完成声明但缺少 [VERIFY] 或 VERDICT token。请运行验证命令后再声明完成。
Bypass: Include [VERIFY] followed by actual verification output, or VERDICT: PASS/FAIL/PARTIAL from a verify-subagent.
Triggers when: Response is >200 characters of code block content (``` fenced) with <30 characters of natural language text outside the fence.
Block message: [no-tool-gate] 纯代码块响应缺少说明。请补充 tool call 或说明下一步操作。
Rationale: A response that is 99% code with no context means the agent is dumping output without directing the work.
Triggers when: last_assistant_message is empty, or contains [max_tokens]/[truncated] markers, or is under 10 characters.
Block message: [no-tool-gate] 响应不完整，请重新生成。
Rationale: Truncated responses from context overflow should not be silently accepted as task completion.
When a block fires, the hook outputs:
{"decision": "block", "reason": "[no-tool-gate] <specific message>"}
Exit code 1. The agent receives the reason as a continue-prompt injection.
[VERIFY] must appear literally, not as a paraphrase.[VERIFY] or VERDICT:.constraints/block-patterns.md for Layer 0 pattern definitions.