dont-make-me-think
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data by auditing third-party codebases and interacting with live URLs using Playwright (Phase 1). This creates an attack surface for indirect prompt injection, where malicious instructions embedded in a target website or codebase could attempt to influence the agent's behavior.
- Ingestion points: Phase 1 (SKILL.md) and reference/audit.md describe reading codebase contents and navigating live URLs.
- Boundary markers: The instructions do not specify explicit delimiters or "ignore instructions" prompts for the external data being audited.
- Capability inventory: The skill possesses the capability to run a browser via Playwright and to modify/commit code to the repository in Phase 3.
- Sanitization: No specific sanitization or filtering of the external audited content is mentioned.
- [COMMAND_EXECUTION]: The skill instructions (Phase 1) require the agent to execute Playwright to walk through journeys in a live browser. While this is an intended function for UX auditing, it involves execution of browser automation tools on potentially untrusted remote content.
Audit Metadata