brainstorming
Pass
Audited by Gen Agent Trust Hub on Jun 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill leverages local commands including directory listing, Grep for keyword searching, and Git log inspection to understand the project structure and history.
- [DATA_EXFILTRATION]: The skill explicitly directs the agent to search for sensitive security-related keywords like 'auth', 'login', 'session', and 'JWT' within the codebase to refine requirements. While used for context gathering, this pattern could lead to the exposure of hardcoded credentials or secrets within the generated brainstorming documents or the agent's interaction history.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its codebase investigation features.
- Ingestion points: Reads various project files (README.md, package.json, pyproject.toml, etc.), grep results, and commit history from the repository.
- Boundary markers: None. The skill does not instruct the agent to use delimiters or treat the content of files as untrusted data separate from the task instructions.
- Capability inventory: The skill has file-writing capabilities (creating markdown notes in the specs/ directory) and access to exploration tools.
- Sanitization: There is no specified logic to sanitize or filter potentially malicious instructions embedded in the project files being analyzed.
Audit Metadata