Learn Cloche
This section is the complete reference for Cloche. Whether you are writing your first .cloche workflow file, wiring together multi-step agent pipelines, or configuring which LLM agent runs inside your containers, everything you need is here.
The guides below cover the DSL that defines workflows, the CLI that drives them, and the agent configuration that controls how work is executed.
| Section | Description |
|---|---|
| Basic Concepts | A tour of the four primitives: workflows, steps, wires, and agents |
| DSL Reference | Complete syntax reference for .cloche workflow files |
| Communicating Between Steps | Passing data via env vars, the KV store, and shared files |
| Intent Continuity | Standing project requirements extracted once and injected into agent prompts |
| CLI Reference | All cloche commands and flags |
| Agent Setup | Configuring Claude Code, Codex, and other agents |
| Configuration Reference | Project and user configuration files |
| Safety Guide | Security best practices for running LLM agents |
| Web Console | Browser console for watching runs, triaging what needs you, and answering agent questions |
| Project Layout | Directory structure of a Cloche project |
| Dockerfile Requirements | Requirements for Cloche container images |
| Tutorial: Automating GitHub Issues | End-to-end pipeline from issue to pull request |