research
Pass
Audited by Gen Agent Trust Hub on Jul 28, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill retrieves research data from diverse external sources, including LinkedIn profiles, G2 reviews, and corporate domains via
swan-fetch-scraped-url. This is a core functionality for the intended purpose of company and persona research. - [COMMAND_EXECUTION]: The agent is instructed to use
swan-execute-codeto process large tool outputs. It generates Python scripts using libraries such aspandas(for data aggregation) andbeautifulsoup4(for HTML parsing) to synthesize retrieved data into structured reports. - [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests and processes untrusted data from external websites. * Ingestion points: Scraping instructions across all sub-pages in the
references/directory. * Boundary markers: Not explicitly defined in instructions. * Capability inventory: Access toswan-execute-codeandswan-update-company. * Sanitization: No explicit sanitization or content filtering is defined beyond the agent's summarization instructions.
Audit Metadata