complete-everything
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill dynamically identifies and executes verification tools such as linters, typecheckers, and test suites by parsing repository files including README.md, AGENTS.md, and CLAUDE.md.- [PROMPT_INJECTION]: The instructions utilize directives like 'Do not stop at phase boundaries' and 'Do not ask whether to continue' to promote agent autonomy and prevent task interruptions.- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface area for indirect injection by executing commands derived from potentially untrusted project files.
- Ingestion points: Manifests, CI configs, and project documentation (Phase 0).
- Boundary markers: None provided to isolate discovered commands from the agent's internal instruction set.
- Capability inventory: High-privilege command execution for building and testing code.
- Sanitization: No validation or escaping of discovered command strings is performed before execution.
Audit Metadata