Documentation.
Set up your project, share instructions across coding tools, and check the generated files.
Set up your projectFind the guide for the job.
New to agnostic-ai? Read Why agnostic-ai?, then Set up your project. Already set up? Pick a guide, or look up a command or field in Reference.
Start here
Why agnostic-ai?Keep one project setup and generate the files each coding tool reads. InstallationInstall agnostic-ai on macOS, Linux, or Windows, then verify and upgrade it. Set up your projectShare one project note across your coding tools, then change it once and update them together. Set up agnostic-ai with a coding agentPaste one prompt into your coding agent and it sets up agnostic-ai for you. Import existing tool configurationImport an existing AI tool setup into agnostic-ai and keep its behavior.
Guides
Directory-specific instructionsKeep rules close to the directories where they apply across supported tools. Local overridesKeep personal specs and instructions on top of the shared ones, out of Git, per project or per machine. Inspect your configurationFind a warning in the browser, edit its source, and sync the fix. CIFail CI when generated files no longer match your specs. Git hooksCheck for out-of-date generated files at commit time, and regenerate them after checkout or pull. Spec packsInstall, update, and share collections of agnostic-ai specs you can reuse. Shared memoryOne project memory that every AI tool reads and writes, so what one tool learns the others know next session. Session handoffsWrite a portable session handoff and resume work in another AI tool on the same machine. Use Composer with Claude Code and CodexKeep one set of project instructions, rules, and skills for Composer's Claude and Codex providers. Use RTK and Caveman with agnostic-aiTurn on RTK for shorter command output and Caveman for shorter answers, and what each one costs.
Reference
Spec formatWhat each folder under .agnostic-ai/ holds, and the rules every spec shares. CLI referenceLook up every agnostic-ai command, option, and exit behavior. Compare targetsSee side by side what agnostic-ai writes for two or three AI coding tools. Cross-target behaviorUnderstand shared instruction files, machine-wide settings, and differences between tools. ConfigurationChoose tools, source folders, generated file paths, and settings for your project or machine. Global configurationInstall and manage your own cross-project instructions, skills, settings, and MCP servers. Verify a releaseCheck that the agnostic-ai binary you install is the one this repository's release workflow built. Error codesFind an AAI error code, its cause, and the command or edit that fixes it. TroubleshootingFix configuration, sync, scoped-rule, and generated-file problems.
Find something else
| You want to | Go to |
|---|---|
| Work in a project that is already set up | Install, run agnostic-ai sync, then follow Daily use |
| See your configuration and warnings in a browser | Inspect your configuration |
| Look up a command, field, or supported tool | CLI reference, Configuration, Supported tools |
| Fix a failed command or missing file | Troubleshooting |
Terms used in these guides
A spec is a source file you edit. A target is a coding tool. Generated files are what sync writes for each tool, and drift means they no longer match their sources.
For coding agents, agent-setup.txt holds the setup checklist. llms.txt lists the guides, and llms-full.txt holds the full documentation.
Contributing to agnostic-ai itself? Use the contributor setup.