skill-forge

Pass

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill provides a framework for creating and reviewing AI agent skills, which involves ingesting untrusted data from user-provided workflows and existing SKILL.md files.
  • Ingestion points: User-supplied text descriptions, conversation history mining, and existing markdown files processed in 'Improve Mode', 'Review Mode', and 'Extract Mode'.
  • Boundary markers: The instructions do not define specific delimiters or 'ignore' instructions for the agent when processing external content, increasing the risk of the agent obeying instructions embedded within the data.
  • Capability inventory: The skill includes local Python scripts (validate_skill.py and package_skill.py) that perform file system operations (read, write, zip) within the provided path. The instructions also suggest the use of MCP tools and code execution for validation.
  • Sanitization: There is no evidence of sanitization or content filtering for the data being ingested and processed by the Skill Forge assistant.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 3, 2026, 11:38 AM