create-agent-skills
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides structured guidelines, templates, and procedures for building agent skills, with all actions and scripts aligned with the stated educational and auditing purposes.\n- [EXTERNAL_DOWNLOADS]: The documentation references and links to official repositories and guidelines from trusted organizations including Anthropic, OpenAI, and Microsoft, as well as established industry resources like agentskills.io.\n- [COMMAND_EXECUTION]: Includes a Python utility script
check_reference_structure.pyfor validating the integrity of skill directory structures. The script uses Python standard libraries and performs safe, non-destructive file operations.\n- [DYNAMIC_EXECUTION]: The evaluation framework (evals/evals.json) employs assertion scripts that execute Python and Ruby code to verify the correctness of agent-generated skill content. These operations are confined to the testing environment and follow standard evaluation protocols.
Audit Metadata