Lindy

Add a TokenMix call to a Lindy agent with the HTTP Request action

What it is

Lindy is a no-code platform for building autonomous AI agents and assistants that handle tasks like inbox triage, scheduling, follow-ups, and CRM updates across thousands of connected apps. You build a flow from triggers and actions.

Before you begin

• A Lindy account.
• A TokenMix account and an API key from tokenmix.ai/dashboard/keys (sk-tm-...).
• The endpoint https://api.tokenmix.ai/v1/chat/completions.

The model that runs a Lindy agent is chosen from Lindy's built-in providers and can't be pointed at a custom endpoint. But you can call TokenMix as a step inside a flow using the HTTP Request action.

Connect TokenMix

1. In the Flow Editor, add a step and choose Perform an action, then HTTP Request.
2. Set the URL to https://api.tokenmix.ai/v1/chat/completions and the method to POST.
3. Add headers: Authorization = Bearer sk-tm-... and Content-Type = application/json.
4. Set the JSON body, for example: {"model":"claude-sonnet-4.6","messages":[{"role":"user","content":"..."}]}
5. Use the returned choices[0].message.content in later steps of the flow.

Verify it works

Run the flow once. A normal response with a choices array means the call reached TokenMix. Confirm the spend in your TokenMix dashboard under usage.

FAQ

Can I change the agent's main model to TokenMix? Not directly. The agent's reasoning model comes from Lindy's built-in providers. The HTTP Request action lets you call TokenMix for specific steps.

Which model should I use? For routine steps, a fast model like deepseek-v4-flash keeps costs low; for harder tasks, use claude-sonnet-4.6 or gpt-5.5.