research-agent

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • Prompt Injection (LOW): Vulnerability to Indirect Prompt Injection (Category 8) identified. The skill ingests untrusted data from external websites via Firecrawl and Perplexity and synthesizes it into reports. \n
  • Ingestion points: External data is ingested via scripts/mcp/firecrawl_scrape.py and scripts/mcp/perplexity_search.py. \n
  • Boundary markers: The skill lacks explicit delimiters or instructions to ignore embedded commands in the scraped content. \n
  • Capability inventory: The agent can execute shell commands via uv run and write files to the local file system. \n
  • Sanitization: There is no evidence of sanitization or filtering of external content before processing. \n- Command Execution (SAFE): The skill uses uv run python to execute local scripts (nia_docs.py, perplexity_search.py, firecrawl_scrape.py). This is a standard pattern for MCP tool integration and is considered safe in this context.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 05:50 PM