agentic-validators
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists of documentation and architectural guidance for implementing validation hooks. It does not contain executable code, scripts, or malicious patterns. Any tools mentioned (like ESLint or Prettier) are standard developer utilities and are cited as examples for the user to implement locally.
- [INDIRECT_PROMPT_INJECTION]: The skill outlines a pattern where agents execute tools on code modified by the agent. The skill explicitly mitigates risks by recommending that validators be deterministic, fast, and local (avoiding network calls), which reduces the surface area for malicious exploitation through data poisoning. Evidence: 1. Ingestion points: Code files modified by the agent (SKILL.md). 2. Boundary markers: Not explicitly required for this design pattern. 3. Capability inventory: Execution of local CLI tools via shell (SKILL.md). 4. Sanitization: Recommendation to remove network calls and ensure determinism (VALIDATOR_CHECKLIST.md).
Audit Metadata