managing-monorepo

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent on a protocol to search for, 'accept', and follow instructions from additional SKILL.md files found in various project and user-level home directories. This architecture allows the agent to ingest and act upon instructions from untrusted files placed in these locations.\n
  • Ingestion points: monorepo-contract.md specifies search paths in project directories (e.g., .claude/skills/, .cursor/skills/) and the user home directory (~/.agents/skills/, ~/.claude/skills/).\n
  • Boundary markers: The instructions do not define boundary markers or 'ignore embedded instructions' warnings for the discovered content.\n
  • Capability inventory: The agent is directed to 'Accept a dependency skill' and 'Follow it by intent', granting it the authority to execute instructions from those files.\n
  • Sanitization: No validation, verification, or sanitization process is described for the discovered files before they are accepted into the agent's context.\n- [EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install dependencies from an external repository using the command npx skills add antfu/skills --skill [pnpm|turborepo]. This involves fetching and potentially executing code from a source outside of the designated trusted organizations.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 04:19 AM
Security Audit — agent-trust-hub — managing-monorepo