code-vulnerability-review
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to analyze untrusted external data (source code and codebase artifacts), which constitutes a vulnerability surface for indirect prompt injection. \n
- Ingestion points: Processes codebases, diffs, pull requests, and configuration files as defined in the workflow in
SKILL.md.\n - Boundary markers: The instructions lack explicit delimiters or warnings to treat the analyzed code as potentially containing adversarial instructions aimed at the agent.\n
- Capability inventory: The skill utilizes
rgfor local file system search and can hand off tasks to other specialized agents or tools viasubagent-orchestration(SKILL.md).\n - Sanitization: No specific content sanitization or validation routines for the ingested code are described before analysis.\n- [EXTERNAL_DOWNLOADS]: The skill references vulnerability data and threat intelligence reports from well-known security research organizations.\n
- Evidence: Contains links to official CISA vulnerability feeds and security landscape reports from Google, Rapid7, and VulnCheck in
references/vulnerability-patterns.md. These are documented as trusted or well-known sources.
Audit Metadata