The Agent Skills Directory

[SAFE]: The skill is designed for administrative and configuration tasks within the Harness platform. It utilizes official Harness MCP server tools (harness_list, harness_get, harness_create, harness_update) for its operations.
[SAFE]: The skill emphasizes interactive requirements gathering and requires explicit user confirmation of generated YAML before any creation or update actions are taken.
[SAFE]: Secure default practices are enforced, such as referencing connectors for authentication rather than hardcoding credentials, and implementing allowedDomains for network isolation of the agent runtime.
[SAFE]: The skill includes instructions to prefix generated UIDs with ca_ to prevent accidental resource collisions and ensures that every referenced input has a default value to prevent runtime errors.

create-agent