scan-codebase
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to scan and read arbitrary code, documentation, schemas, and migrations (SKILL.md). This untrusted content serves as the primary ingestion point. If this content contains malicious instructions (e.g., in code comments or design docs), the agent might process them as instructions rather than data.
- Ingestion points: The skill reads current code, docs, tests-as-code, routes, schemas, migrations, and adjacent design material across the specified path.
- Boundary markers: There are no explicit instructions to use delimiters or ignore instructions embedded within the codebase being scanned.
- Capability inventory: The skill produces findings and writes them to path/findings.md. It also hands off tasks to triage-findings, test-codebase, and resolve-findings, allowing an injection to potentially influence a multi-step workflow.
- Sanitization: No sanitization or filtering of the ingested content is mentioned.
Audit Metadata