Skills
Navigate
skills/inkeep/open-knowledge/open-knowledge-write-skill/Gen Agent Trust Hub

open-knowledge-write-skill

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user-provided workflows and descriptions to generate and install new agent instructions (SKILL.md files and associated scripts).\n
  • Ingestion points: Processes user intent, example workflows, and content from external URLs (e.g., skills.sh).\n
  • Capability inventory: Employs tools like write, edit, and install to create and deploy executable agent guidance across project and global scopes.\n
  • Boundary markers: Includes instructions for the agent to disclose its actions to the user and warns that scripts should only be included after user review.\n
  • Sanitization: Relies on the agent's logic and user verification rather than automated sanitization of the generated instructions.\n- [COMMAND_EXECUTION]: The instructions suggest using npx skills find and npx skills add as fallback commands. The npx utility downloads and executes packages from the npm registry.\n- [EXTERNAL_DOWNLOADS]: The skill facilitates importing external agent skills from the skills.sh marketplace and public GitHub repositories using a dedicated import tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 03:01 AM
Security Audit — agent-trust-hub — open-knowledge-write-skill