skill-authoring
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides instructional content for agent skill development without including executable scripts or suspicious commands.
- [EXTERNAL_DOWNLOADS]: References external documentation from trusted domains such as platform.claude.com and agentskills.io for guidance on prompt engineering and specifications. These references are used neutrally and do not involve remote code execution.
- [COMMAND_EXECUTION]: Mentions development-related commands (
skills init,skills-ref validate) as part of a manual workflow, rather than automated execution of dangerous shells. These tools are consistent with the development purpose of the skill. - [METADATA_POISONING]: The metadata fields are used for legitimate tracking of upstream sources and follow the vendor's specified format, pointing to well-known repositories.
Audit Metadata