deep-research-offensive
Pass
Audited by Gen Agent Trust Hub on Apr 9, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches vulnerability and threat intelligence data from well-known authoritative sources such as NIST (nvd.nist.gov), CISA (cisa.gov), and MITRE (cve.mitre.org).
- [COMMAND_EXECUTION]: The skill utilizes the Playwright MCP tool to execute JavaScript for extracting data from complex, dynamically rendered websites like Exploit-DB and Twitter proxies.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes and summarizes content from untrusted external sources. * Ingestion points: Data enters the agent context through search results and web scraping tools (Tavily, Firecrawl, and Playwright). * Boundary markers: The workflow uses structured markdown templates and JSON schemas for data extraction to provide structural context to retrieved information. * Capability inventory: The skill is capable of automated web searches, full-page content scraping, and browser-based data retrieval. * Sanitization: No explicit sanitization of scraped text for potential malicious instructions is detailed.
Audit Metadata