tdd-full-coverage
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions include the use of shell commands such as
pnpm testfor running tests and coverage checks,pnpm migratefor database migrations, anddocker-compose up -dfor starting local services. These are standard developer tools and do not involve high-risk parameters or unauthorized access. - [INDIRECT_PROMPT_INJECTION]: The skill describes a workflow for integration testing that requires the agent to interact with and interpret data from local services like databases, caches, and message queues. This creates a surface where external data enters the agent's context.
- Ingestion points: Output from local databases (e.g., Postgres), caches (e.g., Redis), and message queues (e.g., Pub/Sub) during integration tests described in SKILL.md.
- Boundary markers: None specified for isolating service outputs.
- Capability inventory: Execution of terminal commands and file system operations for code development.
- Sanitization: None; the agent is expected to analyze raw test outputs directly.
Audit Metadata