The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes instructions for running project-specific maintenance and validation commands, such as npm run harness:sync-runtime-skills, node --test tools/skill-contracts.test.mjs, and ha agent validate. These are standard development utility commands intended for verifying skill structure and deployment.
[DATA_EXPOSURE]: The documentation outlines the filesystem hierarchy used by the agent to discover skills (e.g., ~/.agents/skills, <project>/.claude/skills). This provides necessary architectural context for developers to place their skill files correctly and does not involve the exfiltration of sensitive information.
[INDIRECT_PROMPT_INJECTION]: As a tool for generating other skills, it possesses an attack surface for processing untrusted instructional data. However, it includes explicit best practices and a 'What not to do' section that warns against including secrets or absolute paths, mitigating common risks associated with automated content generation.

create-skill