deep-research

Pass

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses shell commands to interact with the system clipboard. It attempts to use pbcopy, xclip, wl-copy, or clip.exe depending on the host operating system.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it incorporates content from local project files and user input into a prompt destined for another AI model.
  • Ingestion points: The skill reads local files such as package.json, requirements.txt, and Cargo.toml to gather project context.
  • Boundary markers: The generated prompt uses Markdown headers (## Objective, ## Context) to structure the data, but it does not include explicit instructions for the receiving model to ignore instructions embedded within the provided context.
  • Capability inventory: The skill can read local files and execute system clipboard commands via the shell.
  • Sanitization: There is no evidence of sanitization or escaping of the content read from local files before it is interpolated into the final research prompt.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 12, 2026, 12:42 AM