review-conventions

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze untrusted external data in the form of code diffs and PR content, which creates a vulnerability surface for indirect prompt injection. 1. Ingestion Point: Triggered by PR reviews or code diff analysis in SKILL.md. 2. Boundary Markers: None defined to isolate the untrusted code content from the agent's internal instructions. 3. Capability Inventory: The agent is instructed to perform local execution of development tools (npm test, npm run build). 4. Sanitization: While the skill instructs the agent to check for security flaws, it does not include internal sanitization for the inputs it processes.
  • [COMMAND_EXECUTION]: The instructions include standard development lifecycle commands (npm test, npm run lint, npm run build) used for verifying the correctness of code changes.
  • [SAFE]: The core content of the skill promotes defensive security practices, such as preventing path traversal and ensuring all API requests use authenticated helpers.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 03:00 AM
Security Audit — agent-trust-hub — review-conventions