chrome-extension

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFEDYNAMIC_EXECUTIONPERSISTENCEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The file evals/files/coupon-finder-popup.html contains the use of eval() to execute dynamic content retrieved from localStorage.
  • [PERSISTENCE]: The file evals/files/tab-time-service-worker.js includes a setInterval loop that calls chrome.runtime.getPlatformInfo() to artificially extend the service worker's lifetime, bypassing default termination policies.
  • [EXTERNAL_DOWNLOADS]: The manifest sample evals/files/coupon-finder-manifest.json and the popup evals/files/coupon-finder-popup.html reference a remote script hosted on the well-known jsDelivr CDN.
  • [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it processes external codebase files. 1. Ingestion points: Extension manifest and script files as described in the review workflow. 2. Boundary markers: Core Rule 1 and the topic router in SKILL.md provide structural boundaries. 3. Capability inventory: Use of chrome.* APIs and network operations via fetch or declarativeNetRequest. 4. Sanitization: Core Rule 12 and the relay pattern in references/network-and-dnr.md emphasize strict validation of external message origins.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 03:14 PM
Security Audit — agent-trust-hub — chrome-extension