research-verification
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill outlines a methodology for gathering and acting upon data from external and local sources, which presents a surface for instructions embedded in that data to influence agent behavior.
- Ingestion points: The workflow uses 'Web search / Firecrawl' and 'Read the local source' to gather information (SKILL.md).
- Boundary markers: There are no instructions provided to treat retrieved content as data only or to ignore embedded commands.
- Capability inventory: The skill explicitly directs the agent to 'Run it' (execute code) and 'Ask the installed tool' (execute shell commands) based on the research findings.
- Sanitization: No sanitization or validation logic is specified for the content retrieved from the web or local files.
- [COMMAND_EXECUTION]: The instructions mandate the use of shell commands and code execution to verify technical facts. Evidence includes directions to run 'Actual runtime output, a test, a repro' and to query tools using flags like '--help', '--version', and 'list' (SKILL.md).
Audit Metadata