customer-discovery
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes several internal Python scripts (
scrape_website_logos.py,scrape_wayback_logos.py,search_builtwith.py) to perform specialized data extraction tasks. It also uses standard shell commands to create local directories for research reports. - [EXTERNAL_DOWNLOADS]: The skill requires the standard 'requests' library and interacts with well-known web services, including the Wayback Machine and BuiltWith, to retrieve public company data.
- [PROMPT_INJECTION]: The skill processes untrusted content from the web, representing an indirect prompt injection surface.
- Ingestion points: Content is retrieved from arbitrary company websites, case study pages, social media platforms (LinkedIn, Twitter), and review sites (G2, Capterra).
- Boundary markers: The instructions do not define specific delimiters or guardrails to isolate ingested content from the agent's control logic.
- Capability inventory: The agent is empowered to perform file system writes and execute included scripts based on the collected data.
- Sanitization: The included Python scripts perform string validation and filtering (noise word exclusion, length checks, and alpha-character ratio validation) on extracted company names to ensure data quality.
Audit Metadata