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

# Pureframe AI Agent Setup Instructions

> Machine-readable setup instructions for AI agents connecting to the Pureframe AI MCP server.

You are helping a user connect their AI coding tool or agent to the Pureframe AI MCP server. It gives you tools to search, list, and manage the user's indexed videos: `search_videos`, `list_collections`, `get_collection`, `get_video`, `get_job_status`, `upload_video`.

## 1. Identify which client you are

Determine which of these you're running as, then jump to its config below. If you can't tell, ask the user.

* Claude Code
* Claude Desktop
* Cursor
* VS Code (GitHub Copilot, Agent mode)
* Codex
* OpenCode

## 2. Get an API key

The user needs a Pureframe AI API key (`pf_...`). Keys can only be created from a signed-in browser session — you cannot create one yourself. Direct the user to **[https://platform.pureframe.ai/settings/integrations](https://platform.pureframe.ai/settings/integrations)** to create one, then wait for them to paste it back to you. A `read_only` key is enough for search and list operations.

## 3. Add the MCP server config

All clients run the same local server via `npx -y @pureframeai/mcp`, configured with one environment variable: `PUREFRAME_API_KEY`.

**Claude Code** — run:

```bash theme={null}
claude mcp add pureframe -- npx -y @pureframeai/mcp
```

Then set `PUREFRAME_API_KEY` in the user's shell profile, or add manually to `~/.claude.json`:

```json theme={null}
{
  "mcpServers": {
    "pureframe": {
      "command": "npx",
      "args": ["-y", "@pureframeai/mcp"],
      "env": { "PUREFRAME_API_KEY": "pf_..." }
    }
  }
}
```

**Claude Desktop** — add to `~/Library/Application Support/Claude/claude_desktop_config.json`:

```json theme={null}
{
  "mcpServers": {
    "pureframe": {
      "command": "npx",
      "args": ["-y", "@pureframeai/mcp"],
      "env": { "PUREFRAME_API_KEY": "pf_..." }
    }
  }
}
```

**Cursor** — add to `.cursor/mcp.json`:

```json theme={null}
{
  "mcpServers": {
    "pureframe": {
      "command": "npx",
      "args": ["-y", "@pureframeai/mcp"],
      "env": { "PUREFRAME_API_KEY": "pf_..." }
    }
  }
}
```

**VS Code** — add to `.vscode/mcp.json`. Note the root key is `servers`, not `mcpServers`:

```json theme={null}
{
  "servers": {
    "pureframe": {
      "command": "npx",
      "args": ["-y", "@pureframeai/mcp"],
      "env": { "PUREFRAME_API_KEY": "pf_..." }
    }
  }
}
```

Open Copilot Chat and switch to Agent mode to use it.

**Codex** — add to `~/.codex/config.toml`:

```toml theme={null}
[mcp_servers.pureframe]
command = "npx"
args = ["-y", "@pureframeai/mcp"]

[mcp_servers.pureframe.env]
PUREFRAME_API_KEY = "pf_..."
```

**OpenCode** — add to `opencode.json`:

```json theme={null}
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "pureframe": {
      "type": "local",
      "command": ["npx", "-y", "@pureframeai/mcp"],
      "environment": { "PUREFRAME_API_KEY": "pf_..." }
    }
  }
}
```

## 4. Confirm

Restart or reload the client if needed, then confirm the connection by making a sample tool call — e.g. search the user's videos for something they mention, or list their collections.

Full per-client reference: [https://docs.pureframe.ai/agents/connect-an-agent](https://docs.pureframe.ai/agents/connect-an-agent)