### Install and Run Mint CLI

Source: https://github.com/mnfst/docs/blob/main/README.md

Install the Mint CLI globally and then run it to start the local development server. Open http://localhost:3000 in your browser.

```bash
npm i -g mint
mint dev
```

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

### Quick Install Script

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Execute this command to download and run the Manifest installer script. It automates the setup process for a self-hosted instance.

```bash
bash <(curl -sSL https://raw.githubusercontent.com/mnfst/manifest/main/docker/install.sh)
```

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

### Start Ollama Server

Source: https://github.com/mnfst/docs/blob/main/providers/local-models.mdx

Pull a model and start the Ollama server. Ensure Ollama is installed and accessible.

```bash
ollama pull llama3.1:8b   # then:
ollama serve
```

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

### Install Mintlify Skill

Source: https://github.com/mnfst/docs/blob/main/AGENTS.md

Installs the Mintlify skill for product knowledge. Run this command in your terminal.

```bash
npx skills add https://mintlify.com/docs
```

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

### Review and Run Installer Script

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Download the installer script first to review its contents before execution. This allows for manual inspection and execution.

```bash
curl -sSLO https://raw.githubusercontent.com/mnfst/manifest/main/docker/install.sh
less install.sh
bash install.sh
```

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

### Install Mintlify CLI

Source: https://github.com/mnfst/docs/blob/main/CONTRIBUTING.md

Install the Mintlify CLI globally using npm. This is a prerequisite for local documentation development.

```bash
npm i -g mint
```

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

### Start LM Studio Server

Source: https://github.com/mnfst/docs/blob/main/providers/local-models.mdx

Start the LM Studio local inference server. This can be done via the command line or through the LM Studio application.

```bash
lms server start
```

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

### Start Local Development Server

Source: https://github.com/mnfst/docs/blob/main/CONTRIBUTING.md

Navigate to the docs directory and run the Mintlify development server. This allows you to preview changes locally before submitting a pull request.

```bash
mint dev
```

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

### Start llama.cpp Server on Host 0.0.0.0

Source: https://github.com/mnfst/docs/blob/main/providers/local-models.mdx

Start the llama.cpp server, explicitly binding to all network interfaces (0.0.0.0) and specifying the port. This is necessary for Docker compatibility.

```bash
llama-server -m <your-model>.gguf --host 0.0.0.0 --port 8080
```

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

### Start llama.cpp Server

Source: https://github.com/mnfst/docs/blob/main/providers/local-models.mdx

Launch the llama.cpp server with a specified GGUF model file and port. Replace <your-model>.gguf with the actual model path.

```bash
llama-server -m <your-model>.gguf --port 8080
```

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

### Start LM Studio Server on Local Network

Source: https://github.com/mnfst/docs/blob/main/providers/local-models.mdx

Start the LM Studio server, binding to all network interfaces and enabling CORS. This is useful when running Manifest in Docker.

```bash
lms server start --bind 0.0.0.0 --port 1234 --cors
```

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

### Upgrade Manifest Installation

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Upgrade an existing self-hosted Manifest installation to the latest release. This involves pulling the new image and restarting the services.

```bash
docker compose pull
docker compose up -d
```

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

### Run Manifest with BYO PostgreSQL (Linux/macOS)

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Start the Manifest Docker container, connecting it to an existing PostgreSQL database. Ensure you replace placeholders with your actual credentials.

```bash
docker run -d \
  -p 2099:2099 \
  -e DATABASE_URL=postgresql://user:pass@host:5432/manifest \
  -e BETTER_AUTH_SECRET=$(openssl rand -hex 32) \
  -e BETTER_AUTH_URL=http://localhost:2099 \
  manifestdotbuild/manifest
```

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

### Restore Manifest Database

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Restore the Manifest database from a SQL backup file. This involves starting the postgres service, piping the backup file into the psql command, and then starting the rest of the stack.

```bash
docker compose up -d postgres
cat manifest-backup.sql | docker compose exec -T postgres psql -U manifest manifest
docker compose up -d
```

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

### Start Manifest Stack with Docker Compose

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Launch the Manifest services in detached mode using Docker Compose after configuring the environment variables.

```bash
docker compose up -d
```

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

### Run Manifest with BYO PostgreSQL (Windows CMD)

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Start the Manifest Docker container on Windows using the Command Prompt, connecting to an external PostgreSQL database. You must provide your own generated secret.

```cmd
docker run -d ^
  -p 2099:2099 ^
  -e DATABASE_URL=postgresql://user:pass@host:5432/manifest ^
  -e BETTER_AUTH_SECRET=<your-64-char-secret> ^
  -e BETTER_AUTH_URL=http://localhost:2099 ^
  manifestdotbuild/manifest
```

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

### Run Manifest with BYO PostgreSQL (Windows PowerShell)

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Start the Manifest Docker container on Windows using PowerShell, connecting to an external PostgreSQL database. A random secret is generated dynamically.

```powershell
$secret = -join ((48..57 + 97..122) | Get-Random -Count 64 | ForEach-Object { [char]$_ })

docker run -d `
  -p 2099:2099 `
  -e DATABASE_URL=postgresql://user:pass@host:5432/manifest `
  -e BETTER_AUTH_SECRET=$secret `
  -e BETTER_AUTH_URL=http://localhost:2099 `
  manifestdotbuild/manifest
```

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

### Example Error Message

Source: https://github.com/mnfst/docs/blob/main/errors/M001.mdx

This is the typical error message displayed when the M001 error occurs. It includes a link to the documentation for further details.

```text
[🦚 Manifest M001] Missing the Authorization header. Set it to "Bearer mnfst_<your-key>".
See https://manifest.build/docs/errors/M001
```

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

### Example Manifest Error Message

Source: https://github.com/mnfst/docs/blob/main/errors.mdx

This is how a Manifest error message typically appears in a response, including the error code and a link to more details.

```text
[🦚 Manifest M100] No anthropic API key yet. Add one here: https://app.manifest.build/...
See https://manifest.build/docs/errors/M100
```

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

### Manifest M003 Error Message

Source: https://github.com/mnfst/docs/blob/main/errors/M003.mdx

This is the error message displayed when the Manifest key format is invalid. It indicates that the key should start with 'mnfst_' and provides a link to the documentation for more details.

```text
[🦚 Manifest M003] That doesn't look right. Manifest keys start with "mnfst_". Grab yours from the dashboard.
See https://manifest.build/docs/errors/M003
```

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

### HTTP Authentication Header

Source: https://github.com/mnfst/docs/blob/main/reference/api.mdx

All requests to the Manifest API require a Manifest agent key in the Authorization header. Generate keys from the dashboard's Agents page; they always start with 'mnfst_'.

```http
Authorization: Bearer mnfst_YOUR_KEY_HERE
```

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

### Download Docker Compose and Environment File

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Manually download the Docker Compose configuration and environment variable template file for self-hosting Manifest.

```bash
curl -O https://raw.githubusercontent.com/mnfst/manifest/main/docker/docker-compose.yml
curl -O https://raw.githubusercontent.com/mnfst/manifest/main/docker/.env.example
cp .env.example .env
```

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

### Backup Manifest Database

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Create a SQL backup of the Manifest database while the stack is running. This command connects to the postgres service and dumps the 'manifest' database.

```bash
docker compose exec -T postgres pg_dump -U manifest manifest > manifest-backup-$(date +%F).sql
```

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

### Verify Manifest API Connection

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Send a test request to the Manifest API to verify your connection and API key. Ensure you replace the placeholder key with your actual Manifest API key.

```bash
curl -X POST http://localhost:2099/v1/chat/completions \
  -H "Authorization: Bearer mnfst_YOUR_KEY_HERE" \
  -H "Content-Type: application/json" \
  -d '{"model": "manifest/auto", "messages": [{"role": "user", "content": "Hello"}]}'
```

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

### Verify Image Signature with Cosign

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Verify the signature of a published Manifest image before pulling it. This uses cosign keyless signing and requires specifying the image and certificate details.

```bash
cosign verify manifestdotbuild/manifest:<version> \
  --certificate-identity-regexp="^https://github.com/mnfst/manifest/" \
  --certificate-oidc-issuer="https://token.actions.githubusercontent.com"
```

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

### Check for Broken Links

Source: https://github.com/mnfst/docs/blob/main/AGENTS.md

Scans your documentation project for any broken internal or external links. Run this command to ensure link integrity.

```bash
mint broken-links
```

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

### Restart Dev Server

Source: https://github.com/mnfst/docs/blob/main/CLAUDE.md

Reload the dev server on the default port (48721) after modifying doc files. Run this command in the background.

```bash
lsof -ti:48721 | xargs kill -9 2>/dev/null; npx --yes mintlify dev --port 48721
```

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

### Generate Authentication Secret

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Generate a secure, random secret for authentication by running this command. This secret is used in the `.env` file for security.

```bash
openssl rand -hex 32
```

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

### Chat Completions

Source: https://github.com/mnfst/docs/blob/main/reference/api.mdx

Handles chat completion requests using the OpenAI format. This endpoint is compatible with most clients, including the OpenAI SDK, LangChain, Vercel AI SDK, and custom HTTP requests.

```APIDOC
## POST /v1/chat/completions

### Description
Handles chat completion requests using the OpenAI format. This endpoint is compatible with most clients, including the OpenAI SDK, LangChain, Vercel AI SDK, and custom HTTP requests.

### Method
POST

### Endpoint
/v1/chat/completions

### Parameters
#### Request Body
- **model** (string) - Required - The model to use for the chat completion. Use `manifest/auto` for automatic routing.
- **messages** (array) - Required - An array of message objects representing the conversation history.
- **temperature** (number) - Optional - Controls randomness. Lower values make output more focused and deterministic.
- **max_tokens** (integer) - Optional - The maximum number of tokens to generate in the completion.
- **tools** (array) - Optional - A list of tools the model can use.
- **tool_choice** (object) - Optional - Controls how the model selects a tool.
- **response_format** (object) - Optional - Specifies the desired format for the response.
- **stream** (boolean) - Optional - If true, streams the response as Server-Sent Events (SSE).

### Request Example
```json
{
  "model": "manifest/auto",
  "messages": [
    {"role": "user", "content": "What is the capital of France?"}
  ]
}
```

### Response
#### Success Response (200)
- **id** (string) - Unique identifier for the completion.
- **choices** (array) - An array of completion choices.
- **created** (integer) - Timestamp of when the completion was created.
- **model** (string) - The actual model ID used for the completion.
- **object** (string) - The type of object returned (e.g., `chat.completion`).
- **usage** (object) - Information about token usage.

#### Response Example
```json
{
  "id": "chatcmpl-123",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "The capital of France is Paris."
      },
      "logprobs": null,
      "finish_reason": "stop"
    }
  ],
  "created": 1677652288,
  "model": "gpt-3.5-turbo-0125",
  "object": "chat.completion",
  "usage": {
    "prompt_tokens": 10,
    "completion_tokens": 7,
    "total_tokens": 17
  }
}
```
```

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

### OpenAI Chat Completions Request

Source: https://github.com/mnfst/docs/blob/main/reference/api.mdx

Send an OpenAI-shaped request to the /v1/chat/completions endpoint. The 'model' field should be 'manifest/auto', and Manifest will rewrite it to the actual model ID before forwarding. Standard OpenAI fields like temperature, max_tokens, and stream are supported.

```bash
curl -X POST http://localhost:2099/v1/chat/completions \
  -H "Authorization: Bearer mnfst_YOUR_KEY_HERE" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "manifest/auto",
    "messages": [
      {"role": "user", "content": "What is the capital of France?"}
    ]
  }'
```

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

### Manage Manifest Data Volume

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

List or remove the named volume used for Manifest data persistence. Use 'docker compose down -v' with caution as it destroys all data.

```bash
docker volume ls | grep pgdata
```

```bash
docker compose down -v    # destroys all data
```

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

### OpenAI Responses

Source: https://github.com/mnfst/docs/blob/main/reference/api.mdx

Handles requests for OpenAI-compatible responses, specifically for models like Codex, `*-pro`, `o1-pro`, and deep-research models.

```APIDOC
## POST /v1/responses

### Description
Handles requests for OpenAI-compatible responses, specifically for models like Codex, `*-pro`, `o1-pro`, and deep-research models.

### Method
POST

### Endpoint
/v1/responses

### Parameters
#### Request Body
- **model** (string) - Required - The model to use for generating the response. Use `manifest/auto` for automatic routing.
- **messages** (array) - Required - An array of message objects representing the conversation history.
- **max_tokens** (integer) - Optional - The maximum number of tokens to generate in the response.
- **temperature** (number) - Optional - Controls randomness. Lower values make output more focused and deterministic.

### Request Example
```json
{
  "model": "manifest/auto",
  "messages": [
    {"role": "user", "content": "Write a short poem about the sea."}
  ],
  "max_tokens": 150
}
```

### Response
#### Success Response (200)
- **id** (string) - Unique identifier for the response.
- **choices** (array) - An array of response choices.
- **created** (integer) - Timestamp of when the response was created.
- **model** (string) - The actual model ID used for the response.
- **object** (string) - The type of object returned (e.g., `text_completion`).
- **usage** (object) - Information about token usage.

#### Response Example
```json
{
  "id": "resp-abcdef1234567890",
  "choices": [
    {
      "index": 0,
      "text": "The ocean vast, a deep blue hue,\nWith waves that crash and skies so true.\nA salty breeze, a gentle sigh,\nAs seagulls circle in the sky.",
      "logprobs": null,
      "finish_reason": "length"
    }
  ],
  "created": 1677652288,
  "model": "codex-davinci-002",
  "object": "text_completion",
  "usage": {
    "prompt_tokens": 15,
    "completion_tokens": 30,
    "total_tokens": 45
  }
}
```
```

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

### Redirect Telemetry Endpoint

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Configure the TELEMETRY_ENDPOINT environment variable to point to a custom URL for receiving telemetry data. This allows for self-hosted fleet dashboards.

```env
TELEMETRY_ENDPOINT=https://telemetry.mycompany.internal/v1/report
```

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

### Reading Custom Headers with Fetch

Source: https://github.com/mnfst/docs/blob/main/reference/headers.mdx

Use this snippet to make a POST request and extract custom 'x-manifest-*' headers from the response. Header names are case-insensitive but typically lowercased by fetch.

```javascript
const response = await fetch("http://localhost:2099/v1/chat/completions", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.MANIFEST_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "manifest/auto",
    messages: [{ role: "user", content: "Hello" }],
  }),
});

console.log(response.headers.get("x-manifest-tier")); // → "simple"
console.log(response.headers.get("x-manifest-model")); // → "gpt-5-mini"
console.log(response.headers.get("x-manifest-provider")); // → "openai"
```

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

### Configure Custom Port for Manifest

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Change the default port (2099) for Manifest if it's already in use. Update both the Docker port mapping and the BETTER_AUTH_URL environment variable.

```bash
docker run -d \
  -p 8080:2099 \
  -e BETTER_AUTH_URL=http://localhost:8080 \
  ...
```

```yaml
ports:
  - '127.0.0.1:8080:2099'
```

```dotenv
BETTER_AUTH_URL=http://localhost:8080
```

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

### M201 Error Message

Source: https://github.com/mnfst/docs/blob/main/errors/M201.mdx

This is the text output you might see when the M201 error occurs. It includes a brief message and a link to the documentation.

```text
[🦚 Manifest M201] Too many requests — wait a few seconds and retry.
See https://manifest.build/docs/errors/M201
```

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

### Correct Chat Completions Request Body

Source: https://github.com/mnfst/docs/blob/main/errors/M300.mdx

This JSON demonstrates the correct structure for a chat completions API request body, including a valid 'messages' array with at least one entry. Ensure your request body conforms to this shape.

```json
{
  "model": "auto",
  "messages": [
    { "role": "user", "content": "Hello" }
  ]
}
```

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

### Anthropic Messages Request

Source: https://github.com/mnfst/docs/blob/main/reference/api.mdx

Send an Anthropic-shaped request to the /v1/messages endpoint. Use 'manifest/auto' for the model, and specify parameters like max_tokens and messages. The 'anthropic-version' header is also required.

```bash
curl -X POST http://localhost:2099/v1/messages \
  -H "Authorization: Bearer mnfst_YOUR_KEY_HERE" \
  -H "Content-Type: application/json" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "manifest/auto",
    "max_tokens": 1024,
    "messages": [
      {"role": "user", "content": "Hello"}
    ]
  }'
```

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

### Stop Docker Compose Services

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Use these commands to stop the running Docker Compose services. The `-v` flag will also remove associated data volumes.

```bash
docker compose down       # Stop services (keeps data)
```

```bash
docker compose down -v    # Stop and delete all data
```

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

### Manifest M002 Error Message

Source: https://github.com/mnfst/docs/blob/main/errors/M002.mdx

This is the error message displayed by Manifest when an empty Bearer token is detected. It indicates the problem and provides a link to further documentation.

```text
[🦚 Manifest M002] The Bearer token is empty. Paste your Manifest key into it.
See https://manifest.build/docs/errors/M002
```

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

### Disable Telemetry

Source: https://github.com/mnfst/docs/blob/main/self-hosted.mdx

Set the MANIFEST_TELEMETRY_DISABLED environment variable to 1 to disable telemetry. This check occurs before any other telemetry actions.

```env
MANIFEST_TELEMETRY_DISABLED=1
```

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

### Error Response Format

Source: https://github.com/mnfst/docs/blob/main/reference/api.mdx

The Manifest proxy returns errors in a standard JSON envelope. This includes a message, type, and status code. Common codes include 401, 402, 424, and 429.

```json
{
  "error": {
    "message": "Limit exceeded: cost usage ($1.23) exceeds $1.00 per day",
    "type": "limit_exceeded",
    "code": 429
  }
}
```

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

### Hard Limit Trigger Message

Source: https://github.com/mnfst/docs/blob/main/set-limits.mdx

When a hard limit is triggered, the proxy returns a 429 error with this message indicating the exceeded cost.

```text
Limit exceeded: cost usage ($X) exceeds $Y per day
```

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

### Anthropic Messages

Source: https://github.com/mnfst/docs/blob/main/reference/api.mdx

Handles requests using the Anthropic Messages API format. This endpoint is compatible with the Anthropic SDK and other clients that speak the Messages API.

```APIDOC
## POST /v1/messages

### Description
Handles requests using the Anthropic Messages API format. This endpoint is compatible with the Anthropic SDK and other clients that speak the Messages API.

### Method
POST

### Endpoint
/v1/messages

### Parameters
#### Request Body
- **model** (string) - Required - The model to use for the messages. Use `manifest/auto` for automatic routing.
- **max_tokens** (integer) - Required - The maximum number of tokens to generate.
- **messages** (array) - Required - An array of message objects representing the conversation.
- **anthropic_version** (string) - Required - The Anthropic API version to use (e.g., `2023-06-01`).

### Request Example
```json
{
  "model": "manifest/auto",
  "max_tokens": 1024,
  "messages": [
    {"role": "user", "content": "Hello"}
  ]
}
```

### Response
#### Success Response (200)
- **id** (string) - Unique identifier for the message response.
- **type** (string) - The type of object returned (e.g., `message`).
- **role** (string) - The role of the sender (e.g., `assistant`).
- **content** (array) - The content of the message.
- **model** (string) - The actual model ID used for the response.
- **stop_reason** (string) - The reason the model stopped generating tokens.
- **usage** (object) - Information about token usage.

#### Response Example
```json
{
  "id": "msg_0123456789abcdef0123456789abcdef",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "Hello there! How can I help you today?"
    }
  ],
  "model": "claude-3-opus-20240229",
  "stop_reason": "end_turn",
  "usage": {
    "input_tokens": 10,
    "output_tokens": 20
  }
}
```
```

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

### M300 Error Message

Source: https://github.com/mnfst/docs/blob/main/errors/M300.mdx

This is the error message displayed when the M300 error occurs. It indicates that the 'messages' array is required.

```text
[🦚 Manifest M300] `messages` array is required.
See https://manifest.build/docs/errors/M300
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.