> ## Documentation Index
> Fetch the complete documentation index at: https://docs.runagain.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Quickstart

> Create a project, get an API key, and send your first trace.

## 1. Create a project

Sign in at [app.runagain.ai](https://app.runagain.ai) and create a project (or
open the one that was created for you). Each project is an isolated workspace
with its own traces, datasets, experiments, and API keys.

## 2. Generate an API key

In your project, open the onboarding on the **Overview** page (or
**Settings → Create API key**) and click **Generate API key**. Copy the
`sk_…` secret — it's shown once.

<Warning>
  The secret is displayed a single time and never again. Store it in your
  secret manager or `.env` file immediately.
</Warning>

## 3. Point your agent at RunAgain

Pick your stack:

<CardGroup cols={3}>
  <Card title="Claude Agent SDK" icon="sparkles" href="/integrations/claude-agent-sdk" />

  <Card title="Vercel AI SDK" icon="triangle" href="/integrations/vercel-ai-sdk" />

  <Card title="Claude Code (CLI)" icon="terminal" href="/integrations/claude-code" />
</CardGroup>

All of them export over OTLP/HTTP to `https://ingest.runagain.ai`, authenticated
with your key via the `x-runagain-key` header.

```bash Environment theme={null}
RUNAGAIN_API_KEY=sk_…
RUNAGAIN_INGEST_URL=https://ingest.runagain.ai
```

## 4. Verify

Produce some activity, then open your project's **Traces** page — your runs
appear within seconds. From there you can inspect spans, score runs, and set up
drift monitors.

<Card title="Not sure where to start?" icon="terminal" href="/integrations/claude-code">
  Tracing Claude Code in your terminal takes three environment variables and no
  code changes.
</Card>