spec-brainstorm
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION] (SAFE): The skill includes explicit 'CRITICAL CONSTRAINTS' that strictly forbid the AI from writing code or editing files, which effectively mitigates risks associated with unintended command execution.
- [EXTERNAL_DOWNLOADS] (SAFE): No references to external package managers (npm, pip) or remote script fetching (curl, wget) were identified.
- [PROMPT_INJECTION] (SAFE): The prompt does not contain any malicious override attempts or bypass instructions. It uses natural instructional language to guide the model's behavior towards a structured design process.
- [DATA_EXFILTRATION] (SAFE): There are no indicators of network communication or attempts to access and transmit sensitive credentials (API keys, SSH keys).
- [INDIRECT_PROMPT_INJECTION] (LOW): The skill allows the agent to check the current project state in the working directory (File: SKILL.md). While this creates an ingestion point for external data, the high-priority instructions to never edit files or write code provide a significant safety barrier against malicious instructions encountered in the filesystem.
Audit Metadata