reviewer
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill uses established development tools (Maven, Gradle, NPM, Composer) and static analysis plugins (SpotBugs, PMD, ESLint, PHPStan) to perform its duties. These are standard practices in software engineering and do not pose a security risk in this context.
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to analyze untrusted source code provided by users. This creates a potential surface for indirect prompt injection if the code contains instructions designed to manipulate the agent. However, the risk is classified as low because the skill is governed by strict, logic-based checklists and predefined reporting templates.
- Ingestion points: Source code files, PR descriptions, and ticket comments from the development workflow.
- Boundary markers: The skill employs structured reporting templates (e.g., Code Review Report) and mandatory severity labels for comments, which help maintain focus on the technical review.
- Capability inventory: Shell command execution for build tools and static analysis, web search/fetch capabilities, and Playwright-based browser automation for UI inspection.
- Sanitization: The skill is explicitly instructed to ignore implementation details in favor of behavioral acceptance criteria and to verify all findings against objective architectural and requirements documents.
Audit Metadata