Build and Operate Kestra with AI
For the complete documentation index, see llms.txt. For a full content snapshot, see llms-full.txt. Append.mdto anykestra.io/docs/*URL for plain Markdown.
Use AI to write, fix, and operate flows, from the Kestra UI or from your own coding agent.
This section is for anyone who builds or runs flows. The AI here helps you do the work and never runs as part of a flow. To call a model, run an AI agent, or expose flows to external agents at runtime, see AI Orchestration.
None of this is required. Every flow you can generate with AI, you can also write in YAML or build in the No-Code editor.
Choose a tool
Start from where you already work:
| Where you work | Use | What it gives you |
|---|---|---|
| The Kestra UI | AI Copilot | Build and edit flows from a prompt, diagnose failed executions, and run approved multi-step plans |
| A coding agent in your terminal or editor, such as Claude Code, Cursor, Codex, or Windsurf | Docs MCP server and Agent Skills | Your agent writes, validates, and deploys flows next to the rest of your codebase |
| Scripts and CI | kestractl | Deterministic flow, execution, and namespace operations with no AI involved |
| Your own AI tool or integration | Docs for Agents | Every machine-readable Kestra resource: llms.txt, Markdown docs, plugin schemas, and OpenAPI specs |
Use your own coding agent
If your flows live in Git next to application code, you can stay in your coding agent. Three pieces work together, and each covers a different gap:
- Docs MCP server for context. A hosted Model Context Protocol endpoint that gives your agent live task schemas for all
0+ plugins, the blueprints library, and documentation search. Without it, an agent guesses plugin properties from stale training data. - Agent Skills for know-how.
SKILL.mdfiles that teach your agent to write valid flows, harden them for production, operate an environment, and migrate from Airflow or to Kestra 2.0. Skills carry the guardrails: no invented task types, no hardcoded secrets, validate before deploying. kestractlto act. The Kestra CLI that the agent calls to validate, deploy, and run flows on your instance. Without it, the agent can write flows but can’t deploy them.
To set up all three, connect the Docs MCP server, install the skills, and configure kestractl. The Agent Skills examples show a single prompt that writes, validates, and deploys a flow.
AI Copilot
Use Kestra AI Copilot to build and edit flows, diagnose failed executions, and run approved multi-step plans from a chat sidebar in the Kestra UI.
Docs MCP Server
Connect Claude Code, Cursor, or another coding agent to the hosted Kestra MCP server for live plugin schemas, blueprints, and docs while you write flows.
Agent Skills
Give AI coding agents like Claude Code, Cursor, and Windsurf structured knowledge to generate Kestra flows and operate Kestra environments using kestractl.
Docs for Agents
Machine-readable resources for AI agents and LLMs: the Docs MCP server, Agent Skills, llms.txt, Markdown docs, plugin schemas, and OpenAPI specs.
Was this page helpful?