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

# Cline

> Configure Cline to use Capriole AI.

## What is Cline

Cline is an open-source coding agent for the terminal and for VS Code. Its OpenAI Compatible provider sends model requests to Capriole AI through Chat Completions.

## Prerequisites

Before you begin, make sure you have:

* A Capriole AI account ([Sign up here](https://capriole.ai))
* A Capriole AI API key ([Get your key](https://capriole.ai?view=api))

## Installation

Install the Cline CLI with npm:

```bash theme={null}
npm install -g cline
```

Verify the install:

```bash theme={null}
cline --version
```

## Configuration

### Model validation

Cline CLI 3.0.73 passes a file-read, file-write, and final-answer check through Capriole AI with every model in [Supported models](#supported-models), using the configuration below.

Save Capriole AI as the provider:

```bash theme={null}
export CAPRIOLE_AI_API_KEY="YOUR_CAPRIOLE_AI_API_KEY"
cline auth -p openai-compatible \
  -k "$CAPRIOLE_AI_API_KEY" \
  -b "https://api.caprioletech.com/v1" \
  -m claude-latest
```

What each setting does:

* `-p openai-compatible` uses Cline's OpenAI Compatible provider, which sends the key as a Bearer token.
* `-b` is the base URL. It includes `/v1`, because Cline adds `/chat/completions` itself.
* `-m` is the default model.

Cline saves these values in `~/.cline`, so you run `cline auth` once.

In the VS Code extension, open Cline's settings, choose **OpenAI Compatible**, and enter the same base URL, API key, and model ID.

## Run Cline

Start Cline from your project directory:

```bash theme={null}
cline
```

For a one-off task, pass the prompt. Use `-m` to pick another model for that run:

```bash theme={null}
cline -m moonshot/kimi-k3 "Explain this repository."
```

## Supported models

All models use Chat Completions.

| Model | Model ID |
| - | - |
| OpenAI latest | `openai-latest` |
| GPT-6.1 Sol | `openai/gpt-6.1-sol` |
| Claude latest | `claude-latest` |
| Opus 5.5 | `anthropic/claude-opus-5-5` |
| Gemini 3.8 Flash | `google/gemini-3.8-flash` |
| Grok 4.7 | `xai/grok-4.7` |
| GLM 5.3 Flash | `zai/glm-5.3-flash` |
| Kimi K3 | `moonshot/kimi-k3` |

`openai-latest` and `claude-latest` always point to our latest flagship in each family. `GET https://api.caprioletech.com/v1/models` lists every model ID.

## Troubleshooting

### Why does Cline return an authentication error?

Run `cline auth` again with your current Capriole AI API key.

### Why does a Claude model fail with `Unsupported chat model_key`?

On Chat Completions, use the `anthropic/`-prefixed Claude IDs, such as `anthropic/claude-opus-5-5`. The bare IDs, such as `claude-opus-5-5`, are for Anthropic Messages.

### Why do requests fail with endpoint errors?

Confirm that the base URL is `https://api.caprioletech.com/v1`.

## Notes

* Cline runs coding tools on your computer. Capriole AI only handles the model requests.
* Official references: [Cline CLI](https://docs.cline.bot/cline-cli/overview) and [Cline OpenAI Compatible provider](https://docs.cline.bot/provider-config/openai-compatible).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.