Documentation
Cumaru gives your coding agents a knowledge tree to navigate. Start with the steps below, in order. Each page solves one task and links its upstream source.
Tutorial Learn by doing, step by step, from start to a working result.
How-to guide Solve one task you already have.
Reference Look up exact commands, fields, and contracts.
Explanation Understand why Cumaru works the way it does.
Start
- Install the CLI Install cumaru with the script, Homebrew, or npm.
- Set up your project Choose a domain and connect Codex or Claude.
- Start new project Agree on project direction and write the first plan.
- Bootstrap existing Map existing code from inspected evidence.
Concepts
- Why Cumaru The problem and the core idea.
- Knowledge tree Nodes, summaries, and guided loading.
- Ownership Framework prose versus your tag bodies.
- Disciplines Execution rules, strictness, and triggers.
Domains
Guides
- Find context Discover knowledge with tree, map, and get.
- Plan to spec Take sdlc-light work from idea to absorbed spec.
- Maintain knowledge Grow specs, references, and summaries.
- Customize Adopter context, targets, and workflows.
- Update and migrate Refresh framework files and cross versions.
- Troubleshooting From symptom to diagnostic to remedy.
Reference
- CLI Every cumaru command.
- Configuration The config.yaml contract.
- Skills Universal, domain, and opt-in skills.
- Agent adapters Files each adapter writes.
- Hooks Session, prompt, and read hooks.
Contributing
Source of truth for every contract: the Cumaru repository.