Loading...
Generate a map from your terminal, then explore it in the browser, your editor, or CI. Pick an on-ramp below.
Choose your on-ramp
Install CodeBoarding with pipx (recommended), then run the one-time setup that downloads the language servers.
Point CodeBoarding at any local repo. It combines static analysis with LLM reasoning and writes the result to .codeboarding/ as analysis.json plus readable markdown docs. No auth, no waiting on CI.
Use codeboarding incremental --local . to re-run only on what changed, --depth-level 2 for a deeper map, or pass a remote URL like codeboarding full https://github.com/owner/repo.
When the run finishes, CodeBoarding opens the webview in your browser automatically with the diagram already loaded, and prints a clickable link to it in your terminal. No upload, no signup; your analysis is served straight from your machine.
--no-open to skip the auto-open. The link is still printed so you can open it whenever you like. You can also load any .codeboarding/analysis.json by hand via “Load a file” in the webview.CodeBoarding works with your preferred LLM provider: OpenAI, Anthropic, Google, Vercel, AWS Bedrock, Ollama, OpenRouter, or a LiteLLM proxy. Set keys via ~/.codeboarding/config.toml or environment variables.
Install the CLI, run it locally, and view the result in the webview. No auth, no waiting. Then bring it into your editor and CI when you're ready.