repo-audit
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run existing lint, typecheck, test, and build commands found within the audited repository. Executing scripts or binaries from an untrusted codebase is a known risk, although the instructions include a 'when safe' qualifier.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data, including repository files, CI configurations, and documentation. This creates an attack surface where a malicious codebase could include instructions (e.g., in a README or AGENTS/instructions directory) intended to manipulate the agent's audit findings or behavior.
- Ingestion points: README, manifests, lockfiles, and repository source files (SKILL.md).
- Boundary markers: None explicitly defined beyond the instruction to 'mark assumptions explicitly'.
- Capability inventory: Execution of shell commands for linting, testing, and building (SKILL.md).
- Sanitization: The skill relies on the agent's judgment ('when safe') rather than automated sanitization or sandboxing.
Audit Metadata