design-mirror
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Executes local shell scripts (scripts/screenshot.sh and scripts/scrape_html.sh) to interact with the Bright Data API using curl. Note: These scripts include the -k flag, which disables SSL certificate verification.
- [EXTERNAL_DOWNLOADS]: Fetches website content, including HTML and screenshots, from user-specified URLs using the vendor's Web Unlocker API (api.brightdata.com).
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external websites, which could contain instructions targeting the agent.
- Ingestion points: Scraped HTML and CSS content is saved to /tmp/target_page.html and analyzed by the agent.
- Boundary markers: The instructions do not define explicit delimiters or instructions to prevent the agent from following directives embedded in the scraped content.
- Capability inventory: The skill can execute local scripts, perform network requests to the vendor API, and modify source files within the user's project.
- Sanitization: No sanitization or validation of the scraped HTML/CSS content is performed before analysis.
Audit Metadata