@quiki/mcp

MCP stdio server for Quiki.io. Exposes the full Quiki toolkit to Claude Desktop / Claude Code / ChatGPT / Gemini / any MCP-capable agent:

• whoami — verify the token and check subscription tier
• scan_url — submit a TikTok, YouTube, or Instagram URL for scanning
• get_scan — poll a previously submitted scan
• list_recent_videos — list the most recently scanned videos in your library
• get_video — fetch the full record (metadata, transcript, AI analysis) for one video
• bulk_import — import a TikTok, YouTube, or Instagram data export
• bulk_import_estimate — preview the credit cost of a bulk import before committing
• get_bulk_import_status — poll a running bulk import

Requires a Quiki API token from a PRO (credit) plan or higher. Generate one in your account under Account > API keys.

Claude Desktop / Claude Code configuration

Add the following to your mcpServers config (e.g. ~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "quiki": {
      "command": "npx",
      "args": ["-y", "@quiki/mcp"],
      "env": { "QUIKI_API_TOKEN": "qk_live_..." }
    }
  }
}

Environment variables

| Variable | Default | Description | |----------|---------|-------------| | QUIKI_API_TOKEN | — | Required. Raw PAT (qk_live_...). | | QUIKI_API_BASE_URL | https://api.quiki.io/api/v1 | Override for staging / local dev (e.g. http://localhost:3000/api/v1). |

Development

npm install          # from repo root
npm run build        # from packages/mcp
npm test             # from packages/mcp