debugging-methodology
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- Indirect Prompt Injection (HIGH): The skill describes methods for reading and interacting with external web pages using Playwright (e.g., page.getByText). This is a Tier HIGH surface because the agent combines external content ingestion with the capability to execute commands (npx playwright). Malicious web content could potentially inject instructions into the agent's workflow. (1) Ingestion points: references/playwright-e2e-debugging.md (page selectors). (2) Boundary markers: None mentioned. (3) Capability inventory: references/playwright-e2e-debugging.md (command line execution). (4) Sanitization: None.
- Unverifiable Dependencies (MEDIUM): references/root-cause-tracing.md refers to a script find-polluter.sh supposed to be in the directory, but it is not provided. Executing unverified scripts is a security risk.
Recommendations
- AI detected serious security threats
Audit Metadata