learn
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to search and ingest the user's codebase to find examples of security vulnerabilities (e.g., SQL injection, authentication flaws) and present them back to the user for educational purposes.
- Ingestion points: The skill uses Glob and Grep tools to read files across the user's codebase (Step 2: Show Code from the User's Codebase).
- Boundary markers: The instructions do not specify any explicit boundary markers or 'ignore embedded instructions' warnings when reflecting user code back into the prompt.
- Capability inventory: The skill has read-only access to the filesystem via search tools but does not demonstrate write capabilities or network operations in the provided file.
- Sanitization: There is no mention of sanitizing or escaping the content retrieved from the user's files before presenting it in the conversation, which could theoretically allow malicious code comments or strings within the user's own codebase to influence the agent's behavior during the walkthrough.
Audit Metadata