# How Clanker Cloud works

Clanker Cloud runs locally, uses your existing provider credentials and AI keys, routes questions to the right cloud or cluster surfaces, and returns grounded answers or reviewed plans.

## Operating loop

1. Connect existing cloud accounts, kubeconfig contexts, repos, and AI keys.
2. Route the request to the relevant provider or local MCP surface.
3. Gather live evidence such as resource state, logs, events, topology, cost, or deploy context.
4. Synthesize the result into an answer, summary, or plan.
5. Review the proposed impact.
6. Apply only when maker mode is intentionally approved.

## Why the model is local-first

- Credentials stay on the operator machine.
- Bring-your-own AI keys keep pricing and provider choice under team control.
- The app exposes a local MCP endpoint for other agents.
- The same core engine is inspectable through the public Clanker CLI.

## Related pages

- [What is Clanker Cloud?](https://clankercloud.ai/what-is-clanker-cloud)
- [Security and trust boundary](https://clankercloud.ai/security)
- [Pricing](https://clankercloud.ai/pricing)
- [Clanker Cloud vs manual cloud consoles](https://clankercloud.ai/clanker-cloud-vs-manual-cloud-consoles)
- [Clanker Cloud vs point tools](https://clankercloud.ai/clanker-cloud-vs-point-tools)