how
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from the local codebase (source code and comments) to generate explanations.
- Ingestion points: Reads files and symbols from the local repository based on the user's question, as defined in
SKILL.mdandreferences/explorer-brief.md. - Boundary markers: None explicitly defined in the explorer brief to distinguish between code implementation and potential embedded instructions in comments.
- Capability inventory: Limited to searching and reading local files via subagents; no file-write, network, or arbitrary command execution capabilities are present across the scripts.
- Sanitization: None; the skill relies on the model's ability to distinguish implementation facts from prose during its synthesis phase.
Audit Metadata