The Agent Skills Directory

[SAFE]: The skill serves exclusively as a conceptual router and instructional reference. It provides high-level descriptions of other potential skills and best practices for session management (such as phase boundaries and context hygiene) without implementing any functional logic that interacts with the system or network.
[SAFE]: Configuration settings in both the YAML frontmatter and the agents/openai.yaml file enforce a restrictive policy. Specifically, disable-model-invocation: true prevents the agent from invoking tools or secondary models, significantly reducing the attack surface.
[SAFE]: The skill mentions external documentation (e.g., aihero.dev) and local file paths (e.g., CONTEXT.md, .env) solely for instructional purposes. There is no evidence of automated data exfiltration, credential harvesting, or remote code execution. The description of the /wizard skill indicates it is a human-in-the-loop process, not an automated agent capability.

workflow-router