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

# Choose a client tool

> Select and configure the right NexusAPI client without mixing protocols or credentials.

Choose the tool for the way you work, then use its dedicated guide. Do not copy Base URLs or configuration files blindly between clients: some clients use OpenAI Chat Completions, while others use native Messages or Responses protocols.

## Quick choice

| What you want to do                        | Recommended guide                                      | Best for                               |
| ------------------------------------------ | ------------------------------------------------------ | -------------------------------------- |
| Work with an AI coding agent in a terminal | [Claude Code](/en/tools/claude-code)                   | Claude-style coding workflows          |
| Use OpenAI Codex in a terminal             | [Codex CLI](/en/tools/codex-cli)                       | Responses / Codex workflows            |
| Manage more than one local CLI provider    | [CC Switch](/en/tools/cc-switch)                       | Switching between saved configurations |
| Use Claude Code in VS Code or Cursor       | [VS Code / Cursor](/en/tools/cursor-vscode)            | Editor-first workflows                 |
| Build your own script or application       | [Integration overview](/en/guide/integration-overview) | SDKs and automation                    |

## Before changing a local configuration

1. Create an API key and set a suitable project limit.
2. Confirm that its token group includes the target model.
3. Copy the exact model ID from the Model Marketplace or `/v1/models`.
4. Back up an existing configuration file before editing it.
5. Test with one small, read-only request and check the usage log.

<Warning>
  **Do not run unknown one-click scripts**

  Installation or configuration scripts can modify local settings, write credentials, or install third-party code. Use official sources, read a script before running it, and never paste an API key into a script you do not trust.
</Warning>

## Match the tool to the protocol

| Client type                             | Typical Base URL           | Start with                    |
| --------------------------------------- | -------------------------- | ----------------------------- |
| OpenAI-compatible SDK or custom program | `https://nexusapi.link/v1` | `POST /v1/chat/completions`   |
| Claude Code                             | `https://nexusapi.link`    | Native Messages configuration |
| Codex CLI                               | `https://nexusapi.link/v1` | Custom Responses provider     |

The available protocol depends on the model and token group. Read the [API and model capability matrix](/en/developer/capability-matrix) before deciding that a visible model should work in a specific client.

## A tool is not listed here

If the client supports a custom OpenAI-compatible Base URL, begin with the [quick start](/en/guide/quick-start) and that client's official documentation. Confirm whether it adds `/v1` automatically, which protocol it sends, and whether it permits a custom model ID. We do not publish unverified client instructions merely to expand the list.