release-openclaw-ci
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Node.js scripts (e.g.,
verify-provider-secrets.mjs) and GitHub CLI (gh) commands to perform release preflights, monitor workflow runs, and triage failures. These actions are standard for a CI/CD management tool and are well-defined within the instructions. - [DATA_EXFILTRATION]: The
verify-provider-secrets.mjsscript performs network requests to external AI provider endpoints including OpenAI, Anthropic, Fireworks, and OpenRouter. These requests are used to verify API key validity before starting expensive CI matrices. These are well-known and trusted services, and the script is specifically designed to output only status codes rather than the secret values themselves. - [CREDENTIALS_UNSAFE]: The skill demonstrates high security awareness regarding credentials. It mandates the use of secret management tools like 1Password and provides instructions to avoid environment dumps, secret output in logs, and overwriting valid secrets with unvalidated candidates. The preflight script probes API endpoints to ensure tokens are billable/active without exposing them.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data such as GitHub run logs and PR metadata. While this represents a potential ingestion surface for untrusted data, the instructions provide structured triage steps and specific validation logic (e.g., verifying PR fallback schemas) to mitigate risks associated with processing external CI state.
Audit Metadata