diagnose-ppr

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Employs standard shell utilities such as grep and awk, along with agent-browser commands, to verify project configuration and automate diagnostic workflows.
  • [SAFE]: The skill's operations—specifically using a temporary cookie lock and capturing screenshots—are legitimate techniques for debugging web application performance.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it processes untrusted data from the web pages it analyzes.
  • Ingestion points: Browser snapshots and suspense boundary metadata are saved to /tmp/diagnose-ppr-shell-snap.txt and /tmp/diagnose-ppr-shell-suspense.txt for grading.
  • Boundary markers: The instructions do not define clear delimiters or use 'ignore' warnings for the data captured from the browser.
  • Capability inventory: The skill can navigate the browser, set cookies, execute Javascript via eval, and read local configuration files.
  • Sanitization: Data retrieved from the browser is not sanitized or validated before being analyzed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 08:26 PM
Security Audit — agent-trust-hub — diagnose-ppr