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

# CrewAI

> Route every CrewAI agent call through Toolken. See cost, tokens, and latency per agent in your dashboard with no changes to your crew logic.

**Two lines of config and your entire crew is tracked.** Pass a `LLM` object built with Toolken's base URL and headers to each `Agent`, and every call is attributed and visible in your dashboard -- no monkey-patching, no plugins.

<Steps>
  <Step title="Get your two keys">
    You need two keys:

    | Key          | Where to get it                                     | Looks like    |
    | ------------ | --------------------------------------------------- | ------------- |
    | Toolken key  | Dashboard, under API Keys                           | `tk_live_...` |
    | Provider key | Your OpenAI account (or whichever provider you use) | `sk-...`      |

    Toolken forwards your provider key to the upstream model provider untouched and never stores it.
  </Step>

  <Step title="Build the LLM object">
    Use CrewAI's built-in `LLM` class. Set `base_url` to Toolken's gateway and pass your Toolken key and agent name in `extra_headers`:

    ```python theme={null}
    from crewai import LLM

    llm = LLM(
        model="gpt-4o-mini",
        base_url="https://gateway.toolken.ai/v1",
        api_key="sk-...",          # your provider key (BYOK), forwarded untouched
        extra_headers={
            "X-Toolken-Key": "tk_live_...",
            "X-Toolken-Metadata-Agent": "research-agent",
        },
    )
    ```

    `X-Toolken-Metadata-Agent` is the metadata key that groups this agent's spend in the dashboard. Use any string that is meaningful to you.
  </Step>

  <Step title="Pass the LLM to your agents">
    Hand the `llm` object to each `Agent` you create:

    ```python theme={null}
    from crewai import Agent

    researcher = Agent(
        role="Research Analyst",
        goal="Find and summarize the latest AI research.",
        backstory="You are an expert at reading papers and extracting key findings.",
        llm=llm,
    )
    ```

    No other changes to your `Crew`, `Task`, or `Process` setup are needed.
  </Step>

  <Step title="Confirm in your dashboard">
    Run your crew. Within seconds, calls appear in the Toolken dashboard grouped under the `research-agent` agent. You can see cost, token usage, and latency for every request the agent made.
  </Step>
</Steps>

<Tip>
  Give each agent its own `LLM` instance with a distinct `X-Toolken-Metadata-Agent` value to split spend per agent in the dashboard. For example, define one `LLM` for your researcher with `"X-Toolken-Metadata-Agent": "research-agent"` and a separate one for your writer with `"X-Toolken-Metadata-Agent": "writer-agent"`. The dashboard breaks cost and tokens down by agent so you can see exactly which agent is driving spend.
</Tip>

## Next

<CardGroup cols={2}>
  <Card title="LangGraph" icon="diagram-project" href="/integrations/langgraph">Connect LangGraph nodes to Toolken with a single model config.</Card>
  <Card title="Custom Properties" icon="tag" href="/features/observability/custom-properties">Tag requests with feature, customer, and metadata headers to slice every metric.</Card>
  <Card title="Providers & Routing" icon="route" href="/features/providers-routing/overview">One gateway URL, many providers, your own keys.</Card>
</CardGroup>