grep-domain-expert

Pass

Audited by Gen Agent Trust Hub on Aug 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes several internal Node.js scripts (auth.js, grep-api.js, update-check.js) to perform authentication, API interaction, and update checks. It calculates script paths dynamically using readlink -f to resolve the installation directory.
  • [PROMPT_INJECTION]: The skill presents research reports fetched from an external API, which creates a surface for indirect prompt injection if the source content contains malicious instructions.
  • Ingestion points: Research results returned from the Grep API and displayed to the user in Step 6.
  • Boundary markers: None identified in the presentation instructions.
  • Capability inventory: Execution of shell scripts via Node.js and file system reading.
  • Sanitization: No sanitization or validation of the API-returned content is specified.
  • [REMOTE_CODE_EXECUTION]: The skill uses node -e to dynamically evaluate a JavaScript snippet for parsing JSON job identifiers from the output of the submission command.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 14, 2026, 09:40 PM
Security Audit — agent-trust-hub — grep-domain-expert