grill-me

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill captures raw user input during the interview and writes it to a file (brainstorms/*.md) at the project root. This file is intended to be used as input for other tools (e.g., /pre-implement), creating a surface where malicious instructions provided by a user could influence subsequent agent actions. \n
  • Ingestion points: User responses during the interview process defined in the 'Interview Method' section. \n
  • Boundary markers: The skill uses markdown structure (headers and list items) but lacks explicit sanitization or instructions to ignore embedded commands within the captured answers. \n
  • Capability inventory: File system write access to create the brainstorms/ directory and update markdown files. \n
  • Sanitization: No sanitization or validation of the content of user answers is described before writing to the file. \n- [COMMAND_EXECUTION]: The skill performs file system operations at the project root, specifically creating a brainstorms/ directory and creating/updating markdown files based on the interview session topic and date.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 11:20 PM
Security Audit — agent-trust-hub — grill-me