idea-forge

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes a helper script, scripts/save_to_apple_notes.py, which utilizes the subprocess module to execute the osascript binary. This is a functional requirement to enable interaction with the macOS Apple Notes application for saving generated content via AppleScript.- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection (Category 8) by ingesting untrusted user ideas for analysis. The risk is minimized by the specific domain (idea analysis) and local-only capabilities.
  • Ingestion points: Raw idea text provided as input to the idea trigger defined in SKILL.md.
  • Boundary markers: None identified. User input is processed directly through several R&D interpolation stages.
  • Capability inventory: The skill can perform file system writes via scripts/save_to_obsidian.py and execute AppleScript via scripts/save_to_apple_notes.py.
  • Sanitization: The AppleScript script performs HTML entity escaping on the note body, and the Obsidian script uses a regex-based slugifier to sanitize file names.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 01:07 AM
Security Audit — agent-trust-hub — idea-forge