cover-branches
Pass
Audited by Gen Agent Trust Hub on May 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Executes
git diffin Phase 1 to identify changed source and test files within the repository. - [COMMAND_EXECUTION]: Infers and runs test commands (e.g., via
package.jsonorMakefile) in Phase 3 to verify the generated tests, which involves executing arbitrary shell commands determined by local project configuration. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from source and specification files during Phase 2. Ingestion points: Reads local files identified by git diff or user input in Phase 2. Boundary markers: None identified to separate instructions from code/spec content. Capability inventory: Executes shell commands, writes files, and launches sub-agents. Sanitization: No validation or sanitization of file content is performed before processing.
- [COMMAND_EXECUTION]: Uses the Agent tool in Phase 2 to launch parallel sub-agents for concurrent logic and scenario analysis.
Audit Metadata