pm-agent
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill successfully passes all security checks across the 10 threat categories and follows best practices for task decomposition.- [PROMPT_INJECTION]: No override instructions, jailbreak-style prompts, or instructions to ignore system filters were detected.- [DATA_EXFILTRATION]: The skill does not access sensitive configuration files or hardcoded credentials. It utilizes internal symbol analysis tools for project-scoped architecture reviews, which is consistent with its purpose.- [REMOTE_CODE_EXECUTION]: No patterns for downloading or executing external scripts, binaries, or untrusted code were identified.- [PROMPT_INJECTION]: Evaluation of indirect prompt injection surfaces:
- Ingestion points: Processes user requirements (SKILL.md) and code symbols (error-playbook.md).
- Boundary markers: Not explicitly defined in the provided snippets.
- Capability inventory: Restricted to generating JSON/Markdown plan artifacts; no capabilities for arbitrary command execution or external network calls.
- Sanitization: No explicit sanitization or validation of untrusted input content within the prompt instructions.
Audit Metadata