메뉴
Intercepts irreversible or destructive actions and requires explicit user confirmation before proceeding
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
Handles PR review feedback by fetching comments, grouping issues, fixing one group at a time, and verifying before replies.
Reviews whether a skill will trigger reliably, guide useful behavior, avoid overlap, and produce testable outcomes.
Diagnoses OpenClaw provider, fallback, channel, MCP, and gateway config issues with read-only scans and stateful summaries.
Tracks required validation gates, records pass/fail/waived results, and reports readiness before task completion.
YAML-based delegation grant ledger — issues, validates, and tracks scoped permission grants for sub-agent expansions with token budgets and auto-expiry.
Walks the memory DAG to recall detailed context on demand — query, expand, and assemble cited answers from hierarchical summaries without re-reading raw transcripts.
| name | dangerous-action-guard |
| version | 1.0 |
| category | openclaw-native |
| description | Intercepts irreversible or destructive actions and requires explicit user confirmation before proceeding |
| stateful | true |
Before executing any irreversible or high-impact action, pause and get explicit user confirmation. Log every confirmed and rejected action to an audit trail.
| Category | Examples |
|---|---|
| File destruction | rm -rf, unlink, delete files, empty trash, wipe directories |
| Git destructive | git push --force, git reset --hard, git clean -f, git branch -D |
| External messaging | Send email, post to Slack/Teams/Discord, publish social post, reply-all |
| Financial | Confirm purchase, submit payment, execute trade, cancel subscription |
| Credentials | Rotate/delete API keys, modify OAuth apps, change passwords |
| Infrastructure | Deploy to production, drop database, terminate server instance |
| Permission changes | Share document, change access controls, make resource public |
When about to execute a dangerous action:
Step 1 — Pause before the action
Do not execute the action yet. Write it to pending_action in state with a 5-minute expiry.
Step 2 — Describe to user Tell the user:
Step 3 — Wait for explicit confirmation Accept only unambiguous affirmatives: "yes", "go ahead", "confirmed", "do it", "proceed". Do NOT proceed on: "maybe", "I think so", "sure I guess", or any other hedged response.
Step 4 — Execute within expiry window If confirmed, execute within 5 minutes. If the session lapsed or the user is no longer active, re-confirm.
Step 5 — Log to audit trail Write to state: action, timestamp, user confirmation phrase, outcome (executed / rejected / expired).
Approvals expire after 5 minutes. If you execute a dangerous action more than 5 minutes after receiving confirmation, re-confirm with the user. Stale approvals from prior sessions never carry over.
For bulk operations (e.g. "delete all temp files"), list the specific items and the count before confirming — never confirm a batch without showing scope. If scope exceeds 10 items, show first 5 and the total count.
Every action — confirmed or rejected — is logged to state. Use python3 audit.py --history to review the full trail. The audit trail is the user's safety net for disputed actions.
