ObjectLens MCP Server

A Model Context Protocol (MCP) server that connects Large Language Models (LLMs) to the ObjectLens REST API. This allows AI assistants to browse S3-compatible buckets, search indexed S3 object metadata, and preview object contents directly.

Configuration

The server is configured using environment variables:

Related MCP server: S3 MCP Server

Exposed Tools

The server exposes the following tools to the LLM:

• list_providers: List configured S3 storage providers.

• get_default_provider: Get connection details of the default/active provider.

• list_buckets: List S3 buckets for a specific provider.

• list_bucket_objects: List or search objects in a specific bucket with prefix and pagination.

• get_object_metadata: Retrieve detailed metadata (size, content-type, ETag, etc.) of an object.

• get_object_preview: Read the content/preview of an object (supports text, JSON, CSV, code, etc.).

• search_objects: Query indexed metadata globally or scoped to a bucket.

• scan_bucket: Trigger S3 bucket metadata scanning to sync metadata into ObjectLens database.

• list_activities: Fetch recent activity logs/operations from ObjectLens.

Installation and Run

Run with UV

You can run the server directly using uv:

# From this directory
uv run python server.py

Or run it remotely:

uv run --path /path/to/objectlens/mcp-server/server.py

Docker

Build locally:

Build the Docker image:

docker build -t objectlens-mcp-server .

Run the container:

docker run -i --rm \
  -e OBJECTLENS_API_URL="http://host.docker.internal:8000" \
  objectlens-mcp-server

Pull from GHCR:

The image is automatically built and published to GitHub Container Registry (GHCR) on every push to the main branch or when a release tag (e.g., v1.0.0) is published.

To pull and run the pre-built image directly from GHCR:

docker run -i --rm \
  -e OBJECTLENS_API_URL="http://host.docker.internal:8000" \
  ghcr.io/<github-owner-or-org>/mcp-server:latest

(The -i flag is required because the MCP server communicates over standard input/output).

Integration

Claude Desktop

To integrate this server with Claude Desktop, add it to your configuration file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

• Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "objectlens": {
      "command": "uv",
      "args": [
        "run",
        "--path",
        "/path/to/objectlens/mcp-server/server.py"
      ],
      "env": {
        "OBJECTLENS_API_URL": "http://localhost:8000"
      }
    }
  }
}