workflow-housekeep

Pass

Audited by Gen Agent Trust Hub on Oct 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes various package manager commands (npm, pnpm, yarn, bun, pip, cargo, go, etc.) to audit, update, and verify dependencies. This is a core function of the skill's purpose as a repository maintenance tool.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data from local source files (manifests, READMEs) and external web search results (via Firecrawl) to generate documentation and plan dependency migrations.
  • Ingestion points: Reads package.json, README.md, config files, and web search results for migration guides.
  • Boundary markers: The skill contains internal rules to verify deletions and updates, but does not explicitly use delimiters for third-party search results.
  • Capability inventory: Perform file deletions, file writes (README updates), and package installations.
  • Sanitization: No specific sanitization of external search content is mentioned before it is processed by the agent.
  • [DATA_EXFILTRATION]: The skill identifies and accesses potentially sensitive files such as .env and .env.production to ensure they are not committed to the repository. While this is a security cleanup feature, it involves handling sensitive file paths. The risk is mitigated by the skill's instruction to delete/rotate secrets and update .gitignore.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 7, 2026, 11:27 PM