codebase-summary
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads and processes untrusted codebase content, creating a surface for indirect prompt injection where malicious code could influence the agent's summary.
- Ingestion points: The entire target codebase directory identified in SKILL.md and scanned via prompts in ANALYSIS.md (including source code, package.json, .env, and lockfiles).
- Boundary markers: The research agent prompts in ANALYSIS.md do not utilize delimiters or specific instructions to ignore malicious directives found within the analyzed files.
- Capability inventory: The skill employs agent delegation (Task tool), file system read access for project exploration (ANALYSIS.md), and file system write access for report generation (SKILL.md).
- Sanitization: There is no evidence of sanitization or filtering applied to the codebase content before it is summarized by the AI agents.
- [COMMAND_EXECUTION]: The skill executes a bundled Node.js script to validate the structure of its final output.
- Evidence: ANALYSIS.md Phase 3 includes a step to run
node <skill-dir>/validate-summary.cjson the generated JSON report.
Audit Metadata