keelcode

https://keelcode.ai/docs
Admin
keelcode is a hosted coding agent for your terminal and an OpenAI/Anthropic-grade inference API for...
Tokens:4,061
Snippets:56
Trust Score:3.4
Update:2 months ago
Show doc for...
Tokens:
Raw
### Run guided setup

Source: https://keelcode.ai/docs/cli/commands

Executes the optional guided configuration, such as setting up Nerd Font icons.

```bash
keelcode setup
```

--------------------------------

### Install and initialize the keelcode CLI

Source: https://keelcode.ai/docs

Use the bun package manager to install the CLI globally and authenticate with your account.

```bash
bun add --global @keelcode-ai/keelcode
keelcode login
keelcode
```

--------------------------------

### Install Keelcode CLI globally

Source: https://keelcode.ai/docs/cli/install

Install the package globally using Bun.

```bash
bun add --global @keelcode-ai/keelcode
```

--------------------------------

### Start interactive agent

Source: https://keelcode.ai/docs/cli/commands

Launches the interactive agent for the current or specified project.

```bash
keelcode [project]
```

--------------------------------

### Initialize and use the Keelcode CLI

Source: https://keelcode.ai/docs/quickstart

Install the global package, authenticate, and execute a prompt within a project directory.

```bash
bun add --global @keelcode-ai/keelcode
keelcode login
cd my-project
keelcode -p "find and fix the failing test"
```

--------------------------------

### Start interactive TUI

Source: https://keelcode.ai/docs/cli/usage

Launches the Keelcode agent in the current project directory.

```bash
cd my-project
keelcode
```

--------------------------------

### Authenticate with Keelcode

Source: https://keelcode.ai/docs/cli/install

Commands to initiate the login process and verify the installation.

```bash
keelcode login
keelcode
```

--------------------------------

### List available models via API

Source: https://keelcode.ai/docs/models

Retrieve a list of models available to your account using the GET /v1/models endpoint.

```bash
curl https://api.keelcode.ai/v1/models -H "authorization: Bearer $KEELCODE_API_KEY"
```

--------------------------------

### Model catalog response format

Source: https://keelcode.ai/docs/api/models

Example JSON response containing the list of available models with their IDs and display names.

```json
{
  "data": [
    { "id": "claude-opus-4-8", "type": "model", "display_name": "Claude Opus 4.8" },
    { "id": "gpt-5.6-sol",     "type": "model", "display_name": "GPT-5.6 Sol" },
    { "id": "kimi-k2.6-nitro", "type": "model", "display_name": "Kimi K2.6 Nitro" }
  ]
}
```

--------------------------------

### Streaming response event format

Source: https://keelcode.ai/docs/api/messages

Example of the Anthropic-style server-sent events (SSE) returned by the API during message generation.

```text
event: message_start
data: {"type":"message_start","message":{"id":"msg_...","role":"assistant","model":"claude-opus-4-8","content":[]}}

event: content_block_start
data: {"type":"content_block_start","index":0,"content_block":{"type":"text","text":""}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"Hello"}}

event: content_block_stop
data: {"type":"content_block_stop","index":0}

event: message_delta
data: {"type":"message_delta","delta":{"stop_reason":"end_turn"},"usage":{"output_tokens":12}}

event: message_stop
data: {"type":"message_stop"}
```

--------------------------------

### GET /v1/models

Source: https://keelcode.ai/docs/api/models

Retrieves the catalog of models available to your account. The returned model IDs can be used in the /v1/messages endpoint.

```APIDOC
## GET /v1/models

### Description
List the models available to your account. Responses are cached for 60 seconds.

### Method
GET

### Endpoint
/v1/models

### Response
#### Success Response (200)
- **data** (array) - A list of model objects.
  - **id** (string) - The unique identifier for the model.
  - **type** (string) - The resource type.
  - **display_name** (string) - The human-readable name of the model.

#### Response Example
{
  "data": [
    { "id": "claude-opus-4-8", "type": "model", "display_name": "Claude Opus 4.8" },
    { "id": "gpt-5.6-sol",     "type": "model", "display_name": "GPT-5.6 Sol" },
    { "id": "kimi-k2.6-nitro", "type": "model", "display_name": "Kimi K2.6 Nitro" }
  ]
}
```

--------------------------------

### Authenticate user

Source: https://keelcode.ai/docs/cli/commands

Initiates the device approval sign-in flow.

```bash
keelcode login [--no-browser]
```

--------------------------------

### Authenticate CLI session

Source: https://keelcode.ai/docs/authentication

Use the login command to initiate device approval. Use the --no-browser flag if a browser is unavailable.

```bash
keelcode login
# no browser on this machine?
keelcode login --no-browser
```

--------------------------------

### CLI command aliases

Source: https://keelcode.ai/docs/cli/install

List of available command names and instructions for managing custom aliases.

```text
keelcode
kc
kcode
keel
```

```bash
keelcode alias setup
keelcode alias add mycode
keelcode alias list
```

--------------------------------

### Display account info

Source: https://keelcode.ai/docs/cli/commands

Displays details for the currently authenticated account.

```bash
keelcode whoami
```

--------------------------------

### Run headless prompt

Source: https://keelcode.ai/docs/cli/commands

Executes a single prompt in headless mode, supporting piped stdin.

```bash
keelcode -p "prompt"
```

--------------------------------

### Check credit usage

Source: https://keelcode.ai/docs/cli/commands

Displays current credit consumption.

```bash
keelcode usage
```

--------------------------------

### Manage command aliases

Source: https://keelcode.ai/docs/cli/commands

Provides utilities to add, list, remove, or configure command aliases.

```bash
keelcode alias add|list|remove|setup
```

--------------------------------

### Define tools for model use

Source: https://keelcode.ai/docs/api/messages

JSON body configuration for defining tools with input schemas to enable model-assisted tool calling.

```json
{
  "model": "claude-opus-4-8",
  "max_tokens": 1024,
  "stream": true,
  "tools": [
    {
      "name": "get_weather",
      "description": "Get the current weather for a city.",
      "input_schema": {
        "type": "object",
        "properties": { "city": { "type": "string" } },
        "required": ["city"]
      }
    }
  ],
  "messages": [
    { "role": "user", "content": [{ "type": "text", "text": "Weather in Delhi?" }] }
  ]
}
```

--------------------------------

### List models via cURL

Source: https://keelcode.ai/docs/api/models

Retrieves the list of available models using the Keelcode API key for authentication.

```bash
curl https://api.keelcode.ai/v1/models \
  -H "authorization: Bearer $KEELCODE_API_KEY"
```

--------------------------------

### Execute one-shot prompts

Source: https://keelcode.ai/docs/cli/usage

Runs a single command prompt directly or processes piped input from other tools.

```bash
keelcode -p "find and fix the failing test"
```

```bash
git diff | keelcode -p "review this diff"
```

--------------------------------

### End session

Source: https://keelcode.ai/docs/cli/commands

Terminates the local authentication session.

```bash
keelcode logout
```

--------------------------------

### Create a message with cURL

Source: https://keelcode.ai/docs/api/messages

Sends a POST request to the messages endpoint. Requires an Authorization header with a valid API key and sets stream to true.

```bash
curl https://api.keelcode.ai/v1/messages \
  -H "authorization: Bearer $KEELCODE_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-4-8",
    "max_tokens": 1024,
    "stream": true,
    "messages": [
      { "role": "user", "content": [{ "type": "text", "text": "Hello, keelcode" }] }
    ]
  }'
```

--------------------------------

### Authenticate API requests with Bearer token

Source: https://keelcode.ai/docs/authentication

Include the API key in the Authorization header for all requests.

```bash
Authorization: Bearer $KEELCODE_API_KEY
```

--------------------------------

### Interact with the Keelcode API

Source: https://keelcode.ai/docs/quickstart

Send a streaming request to the Keelcode API using cURL or the Anthropic SDK.

```bash
curl https://api.keelcode.ai/v1/messages \
  -H "authorization: Bearer $KEELCODE_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-4-8",
    "max_tokens": 1024,
    "stream": true,
    "messages": [
      { "role": "user", "content": [{ "type": "text", "text": "Hello, keelcode" }] }
    ]
  }'
```

```ts
import Anthropic from "@anthropic-ai/sdk";

// keelcode speaks the Anthropic Messages API — point the SDK at it.
const client = new Anthropic({
  apiKey: process.env.KEELCODE_API_KEY,
  baseURL: "https://api.keelcode.ai",
});

const stream = await client.messages.stream({
  model: "claude-opus-4-8",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello, keelcode" }],
});

for await (const event of stream) {
  if (event.type === "content_block_delta" && event.delta.type === "text_delta") {
    process.stdout.write(event.delta.text);
  }
}
```

--------------------------------

### POST /v1/messages

Source: https://keelcode.ai/docs/api/messages

Create a message to stream a model response. This endpoint is streaming-only and requires the 'stream' parameter to be set to true.

```APIDOC
## POST /v1/messages

### Description
Stream a model response. This endpoint implements the Anthropic Messages API, allowing for conversational interactions with supported models.

### Method
POST

### Endpoint
/v1/messages

### Parameters
#### Request Body
- **model** (string) - Required - A model id from /v1/models.
- **max_tokens** (integer) - Required - Maximum tokens to generate (1–128000, capped per model).
- **messages** (Message[]) - Required - The conversation. Each message has a role (user | assistant) and an array of content blocks.
- **stream** (true) - Required - Must be true — the endpoint is streaming-only.
- **system** (string | Block[]) - Optional - A system prompt.
- **tools** (Tool[]) - Optional - Tools the model may call.
- **tool_choice** (object) - Optional - auto | any | none | { type: "tool", name }.
- **temperature** (number) - Optional - Sampling temperature, 0–1.
- **top_p** (number) - Optional - Nucleus sampling, 0–1.
- **stop_sequences** (string[]) - Optional - Up to 16 sequences that stop generation.
- **thinking** (object) - Optional - { type: "enabled", budget_tokens } to enable extended thinking.
- **reasoning_effort** (enum) - Optional - low | medium | high — for reasoning models.

### Request Example
{
  "model": "claude-opus-4-8",
  "max_tokens": 1024,
  "stream": true,
  "messages": [
    { "role": "user", "content": [{ "type": "text", "text": "Hello, keelcode" }] }
  ]
}

### Response
#### Success Response (200)
- **SSE Stream** - The response is an Anthropic-style SSE stream containing message_start, content_block_*, message_delta, and message_stop events.
```

--------------------------------

### Status Codes

Source: https://keelcode.ai/docs/errors

List of standard HTTP status codes and their corresponding error codes used by the API.

```APIDOC
## Status Codes

- **400** (invalid_request / invalid_json) - The body is malformed or violates the contract.
- **401** (unauthorized) - Missing or invalid API key.
- **403** (forbidden) - The key lacks the required scope.
- **404** (model_not_found) - Unknown model requested.
- **413** (payload_too_large) - Body exceeds 10 MiB.
- **429** (rate_limited / allowance_exceeded) - Too many requests, or the daily allowance is spent.
- **503** (model_unpriced) - The model is not commercially enabled yet.
```

--------------------------------

### API Error Response Structure

Source: https://keelcode.ai/docs/errors

The standard JSON format for non-2xx API responses.

```json
{
  "type": "error",
  "error": {
    "code": "invalid_request",
    "title": "Invalid request",
    "detail": "The Messages request does not match the API contract."
  }
}
```

--------------------------------

### Error Response Format

Source: https://keelcode.ai/docs/errors

The standard structure for error responses returned by the API.

```APIDOC
## Error Response Format

### Description
Non-2xx responses return a problem object. Clients should branch logic based on the `error.code` field rather than the human-readable `title` or `detail` fields.

### Response Body
- **type** (string) - Always "error"
- **error** (object) - Contains error details
  - **code** (string) - Machine-readable error code
  - **title** (string) - Human-readable summary
  - **detail** (string) - Detailed explanation of the error

### Response Example
{
  "type": "error",
  "error": {
    "code": "invalid_request",
    "title": "Invalid request",
    "detail": "The Messages request does not match the API contract."
  }
}
```
Tokens:
Raw