Skip to content

Quick Start

1. Launch Celebi

After installing, open Celebi from your application launcher or terminal.

2. Setup Wizard

The first time you launch Celebi, you'll see the setup wizard.

Page 1 — Select Project

  • Project Name: A name for this project (e.g., "my-app")
  • Project Folder: Browse to your project directory
  • Agent: Select your coding agent (OpenCode, Claude Code, Codex, or Kiro)

Page 2 — Provider Credentials

  • Provider: Select your LLM provider (Gemini, OpenAI, or Anthropic)
  • API Key: Enter your API key
  • Click Fetch Models to load available models
  • Select a model from the dropdown

3. Start the Proxy

Click Start on the dashboard. Celebi will:

  1. Start a LiteLLM gateway on port 4000
  2. Start the FastAPI proxy on port 8000
  3. Generate the agent config file in your project folder

4. Use Your Agent

Open your coding agent in the project folder as usual. All LLM requests will be transparently captured by Celebi.

5. Inspect Conversations

Use the Celebi dashboard to:

  • Graph: View the conversation flow as a node graph
  • Search: Find specific prompts or responses
  • Analytics: Track token usage and costs

Verification

# Check the proxy is running
curl http://localhost:8000/
# Expected: {"message":"Hello Traveller"}

# Check the graph
curl http://localhost:8000/graph
# Expected: {"nodes":0,"edges":0,"data":[]}