full-cycle
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use standard development and analysis tools, specifically
knip,tsc(TypeScript compiler), andgit. It also calls for executing the repository's native test suites and build processes to validate changes.- [INDIRECT_PROMPT_INJECTION]: The skill describes a process that ingests external code for research ('Peers' stage) and follows repository-specific instruction files likeAGENTS.mdandCLAUDE.md. - Ingestion points: External open-source codebases and local repository configuration/instruction files.
- Boundary markers: Not explicitly mentioned in the skill instructions.
- Capability inventory: Includes file system modification, command execution (tests/builds), and network access (researching peers).
- Sanitization: The skill employs a multi-agent validation strategy where a 'Skeptic' agent must independently reproduce findings before they are finalized, reducing the risk of malicious instructions being blindly followed.
Audit Metadata