Skip to main content
Glama

bucket-mcp

MCP server for Bucket — connect Claude to your Bucket so it can search, read, and download everything you've saved: Instagram reels, TikToks, tweets, YouTube videos, links, and notes.

you: "grab that pasta reel I saved last week and pull the recipe out of the transcript"
claude: searches your bucket → reads the transcript → downloads the video into your project

Setup

1. Get an API key — log in at bucketai.net → Settings → Claude Connector → Generate API Key. Copy it (it's shown once).

2. Add to Claude Code:

claude mcp add bucket -e BUCKET_API_KEY=bk_your_key_here -- npx -y bucket-mcp

Or Claude Desktop — add to claude_desktop_config.json:

{
  "mcpServers": {
    "bucket": {
      "command": "npx",
      "args": ["-y", "bucket-mcp"],
      "env": { "BUCKET_API_KEY": "bk_your_key_here" }
    }
  }
}

Related MCP server: mcp-server-claude

Tools

Tool

What it does

search_bucket

Semantic search over everything you've saved ("that pasta place video")

list_recent

Browse recent saves, filterable by category (instagram, tiktok, twitter, recipes, …)

get_item

Full detail: complete transcript, summary, OCR text, visual description, locations

download_video

Download a saved reel/TikTok/tweet video as an MP4 into your working directory

save_to_bucket

Save a URL into your Bucket from Claude

Configuration

Env var

Required

Default

BUCKET_API_KEY

yes

—

BUCKET_API_URL

no

Bucket production API

Development

npm install
npm run build
BUCKET_API_KEY=bk_... node dist/index.js   # speaks MCP over stdio

License

MIT

Available Tools

5 tools
download_videoA

Download a saved item's video (reel/TikTok/tweet video) as an MP4 file written to the local filesystem. Returns the absolute path of the written file.

Inputs: message_id (required), output_path.

get_itemA

Full detail for one saved item: complete transcript, summary, OCR text, visual description, locations, and original URL.

Inputs: message_id (required).

list_recentA

List recently saved Bucket items, optionally filtered by category (e.g. instagram, tiktok, twitter, youtube, recipes, fitness…).

Inputs: limit, category.

save_to_bucketA

Save a URL (reel, TikTok, tweet, article…) into the user's Bucket. Bucket extracts, summarizes, and categorizes it in the background.

Inputs: url (required).

search_bucketA

Semantic search over everything saved in the user's Bucket (reels, TikToks, tweets, links, notes). Returns matches with summaries; use get_item for full detail.

Inputs: query (required), limit.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A4.1/5.0

    Scored across 5 tools

    Disambiguation5/5

    Each tool serves a distinct purpose: search, list recent, get detail, download video, and save. There is no overlap in actions or resources, making it easy for an agent to select the correct tool.

    Naming Consistency4/5

    Most tools follow a clear verb_noun pattern (search_bucket, get_item, download_video, save_to_bucket), but 'list_recent' omits the noun and 'save_to_bucket' includes a preposition, creating a minor inconsistency. Overall, the style is still predictable and readable.

    Tool Count5/5

    Five tools is well-scoped for a content saving and retrieval service. Each tool covers a distinct operation without unnecessary redundancy, and the count falls comfortably within the ideal range.

    Completeness3/5

    The set covers the main workflow (save, search, list, get detail, download) but lacks a delete operation, which is a fundamental part of managing saved content. This is a notable gap that could force agents to leave items in the bucket indefinitely.

    Maintenance

    ActivityInactive
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      A
      quality
      D
      maintenance
      A Model Context Protocol (MCP) server that enables Claude or other LLMs to fetch content from URLs, supporting HTML, JSON, text, and images with configurable request parameters.
      3
      3
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      An MCP server for storing and retrieving thoughts, notes, and ideas with semantic vector search, enabling integration with Claude Desktop and other MCP clients.
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      Persistent knowledge-base MCP server. Semantic search with source citations over saved YouTube videos, podcasts, articles and PDFs — one endpoint for every MCP-capable AI client.
      28
      Elastic 2.0