Use klaridian from an AI agent

klaridian ships an Agent Skill: a SKILL.md file that teaches a coding agent how to drive the CLI. With it installed, an agent asked to "turn this OpenAPI spec into an MCP server" curates the tools first, passes --json, reads the result's error code instead of guessing from prose, and never passes --interactive without a person at the terminal.

The skill follows the open Agent Skills format, so it isn't tied to one agent.

Install the skill

With the skills CLI, from the repository:

npx skills add klaridian/klaridian

Or from this site, through its discovery index:

npx skills add https://klaridian.dev

Both install the same file. To install it by hand, copy skills/klaridian/SKILL.md into your agent's skills directory, for example ~/.claude/skills/klaridian/SKILL.md for Claude Code.

What the skill covers

• Checking the tool count before generating, and narrowing it with the curation flags or x-klaridian
• The flags agents need most, including language, transport, plugins, and --install
• Reading the JSON output and branching on its error code
• The environment variables a generated server reads for upstream authentication
• Running, inspecting, and deploying the generated server

It doesn't list every flag. For those it points the agent at klaridian generate --help and the CLI reference.

Where it lives

The skill has one source, skills/klaridian/SKILL.md in the repository. This site serves the same bytes at /.well-known/agent-skills/klaridian/SKILL.md and lists it in /.well-known/agent-skills/index.json, following the Agent Skills discovery draft. A test fails the build if the skill mentions a flag, error code, or environment variable that the CLI doesn't have, so it can't drift from the CLI it describes.

For reading the docs rather than driving the CLI, agents can use llms.txt.