VDR Plugins - Video Deep Research

Plugin marketplace for Video Deep Research MCP servers. Search and analyze millions of videos and ads using AI-powered deep research.

Claude Desktop / Cowork (OAuth — recommended)

1. Open the Claude Desktop application.
2. Click Customize from left panel.
3. Click Connectors.
4. Click + and select Add custom connector.
5. Input a connector name and enter the server URL: https://mcp.videodeepresearch.com/mcp

6. Click VDR Connector and click Connect button.

7. Complete the OAuth sign-in flow when prompted. (Make sure you have logged into your Claude.ai account.)
8. The VDR Connector will appear connected.

The setup on claude.ai is similar — go to Settings → Connectors → Add custom connector and enter the same server URL.

No API keys or environment variables needed — Claude handles OAuth automatically.

Claude Code CLI (API key)

1. Add the marketplace

claude plugin marketplace add deepvideolab-ai/vdr-marketplace

2. Install the plugin

claude plugin install vdr-video-research@vdr-plugin

3. Set your API token

Option A: Add to your shell profile (~/.bashrc or ~/.zshrc):

export VDR_MCP_TOKEN='your-api-token-here'

Option B: Edit the installed plugin config at ~/.claude/plugins/vdr-video-research/.mcp.json and replace your-api-token-here with your actual API token:

{
  "mcpServers": {
    "vdr-video-research": {
      "type": "http",
      "url": "https://mcp.videodeepresearch.com/mcp",
      "headers": {
        "Authorization": "Bearer your-api-token-here"
      }
    }
  }
}

4. Use it

Ask Claude Code to search videos:

Search for videos about electric vehicle marketing strategies

Or use the skills directly:

/vdr-video-research:talk-to-1M Find videos about AI coding assistants
/vdr-video-research:talk-to-ads What creative strategies are beauty brands using?

MCP Server (Direct Setup)

If you want to add the VDR MCP server directly without the plugin marketplace, follow these steps.

1. Create an API Key

Go to the VDR dashboard and create a new API key with the required permissions:

https://app.videodeepresearch.com/api-keys

2. Configure Environment Variable

On macOS or Linux, open a terminal and run:

export VDR_MCP_TOKEN='YOUR_API_KEY'

3. Add MCP Server

Claude Code:

Option A: Add VDR_MCP_TOKEN to your shell profile (~/.bashrc or ~/.zshrc), then run:

claude mcp add \
  --transport http \
  vdr-mcp \
  https://mcp.videodeepresearch.com/mcp \
  --header "Authorization: Bearer $VDR_MCP_TOKEN"

Option B: Pass your API token directly in the command:

claude mcp add \
  --transport http \
  vdr-mcp \
  https://mcp.videodeepresearch.com/mcp \
  --header "Authorization: Bearer <your-api-token-here>"

Codex:

Set VDR_MCP_TOKEN in your shell profile (~/.bashrc or ~/.zshrc), then run:

codex mcp add \
  --url https://mcp.videodeepresearch.com/mcp \
  VDR_MCP_TOKEN \
  vdr-mcp

4. Verify MCP Connectivity

Check whether the MCP server has been successfully added:

claude mcp list

or

codex mcp list

OpenClaw Setup

If you use OpenClaw, follow these steps to use VDR video-research via mcporter.

Prerequisites

• A VDR API token (do not paste it into chat)
• mcporter installed:

  npm i -g mcporter

1. Configure mcporter for OpenClaw

OpenClaw uses its own mcporter config at ~/.openclaw/workspace/config/mcporter.json.

Option A: Set the token as an environment variable in your shell profile (~/.bashrc or ~/.zshrc), then reference it in the mcporter command:

export VDR_MCP_TOKEN='your-api-token-here'

mcporter config add vdr-video-research \
  --url "https://mcp.videodeepresearch.com/mcp" \
  --header "Authorization=Bearer $VDR_MCP_TOKEN" \
  --description "Video Deep Research MCP (HTTP, bearer auth)" \
  --persist "$HOME/.openclaw/workspace/config/mcporter.json"

Option B: Pass your API token directly in the mcporter command:

mcporter config add vdr-video-research \
  --url "https://mcp.videodeepresearch.com/mcp" \
  --header "Authorization=Bearer <your-api-token-here>" \
  --description "Video Deep Research MCP (HTTP, bearer auth)" \
  --persist "$HOME/.openclaw/workspace/config/mcporter.json"

If you previously had a broken entry with the same name, remove it first:

mcporter config remove vdr-video-research \
  --persist "$HOME/.openclaw/workspace/config/mcporter.json"

2. Verify the server

mcporter --config "$HOME/.openclaw/workspace/config/mcporter.json" \
  list vdr-video-research --schema

mcporter --config "$HOME/.openclaw/workspace/config/mcporter.json" \
  call vdr-video-research.health --output json

You should see ok: true and tools like talk_to_1m / talk_to_ads.

3. Install the skill

Download dist/vdr-mcporter.skill from this repo and install it:

• OpenClaw app → Settings → Skills → Import / Install from file → select vdr-mcporter.skill
• Or copy manually:

  cp dist/vdr-mcporter.skill ~/.openclaw/workspace/dist/

4. Use it in OpenClaw

Say any of these in the OpenClaw TUI:

• "Use VDR talk-to-1M: electric vehicle marketing strategies"
• "Use VDR talk-to-ads: What creative strategies are beauty brands using?"
• "Run VDR health check"

Common gotchas

• Works elsewhere but not in OpenClaw: You configured a different mcporter.json. Fix by writing the VDR entry into ~/.openclaw/workspace/config/mcporter.json (as above).
• Don't use stdio mode: A stdio vdr-video-research entry that runs mcporter will loop / "connection closed". Use HTTP to https://mcp.videodeepresearch.com/mcp.
• Shell access in TUI: Optional and gated by a one-time prompt per session. Not needed if mcporter config is correct.

Available Plugins

Plugin	Description
vdr-plugin	Search and analyze 5M+ videos and ads with AI-powered deep research

Configuration

Environment Variable	Required	Description
VDR_MCP_TOKEN	Optional and CLI only	Bearer token for API key authentication (not needed for OAuth)

License

MIT