codebase-auditor
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external, untrusted data sources (entire code repositories). An attacker could place malicious instructions within code comments, documentation, or configuration files in the target repo to influence the agent's behavior.
- Ingestion points: The agent is instructed to review "all first-party source code, configuration, infrastructure, schema, migration, test, and documentation files" as specified in the
Repository Coveragesection ofSKILL.md. - Boundary markers: The instructions do not define clear delimiters or provide the agent with specific warnings to disregard instructions found within the analyzed files.
- Capability inventory: The agent has the capability to read files and perform online research/web searches to verify dependency versions.
- Sanitization: There are no sanitization or filtering mechanisms mentioned to handle potentially malicious content in the codebase.
Audit Metadata