fuel-watch
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Node.js scripts (collect.mjs, report.mjs, monitor.mjs) and an external agent-browser CLI tool to perform data scraping. These operations are core to the skill's functionality and are appropriately scoped.
- [INDIRECT_PROMPT_INJECTION]: The skill fetches real-time data from external websites (Yandex, 2GIS, GdeBenz, Benzonavt). Evidence chain: 1. Ingestion points: Scraping results from external mapping and petrol services (e.g., yandex.mjs, twogis.mjs). 2. Boundary markers: Markdown headers and lists (report.mjs). 3. Capabilities: Local command execution for data collection and file system access for state management. 4. Sanitization: Extensive data normalization, brand aliasing, and coordinate verification. The inherent attack surface of scraping is mitigated by structured parsing and is necessary for the skill's primary purpose.
- [EXTERNAL_DOWNLOADS]: The project configuration files and documentation reference standard development dependencies and build tools (npm, esbuild). No untrusted or runtime-executed external downloads were found.
Audit Metadata