do-kb-mcp

noops888/do-kb-mcp

3.3

If you are the rightful owner of do-kb-mcp and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

The Digital Ocean Agent MCP Server is a Model Context Protocol server designed to provide search capabilities for Digital Ocean Agent knowledge bases, enabling AI tools to efficiently retrieve information.

The Digital Ocean Agent MCP Server is a specialized server that facilitates the search and retrieval of information from Digital Ocean Agent instances using the Model Context Protocol (MCP). It supports multiple search methods, including basic, rewrite, step-back, and sub-queries strategies, to optimize the retrieval process. The server is designed to handle multiple agent instances, making it versatile for various configurations. It is deployed using Cloudflare Workers, ensuring scalability and serverless operation. The server incorporates dual-layer rate limiting to protect against excessive costs and is security-hardened with features like sanitized error messages and secure agent selection. By focusing on returning document chunks with minimal AI overhead, the server optimizes for cost efficiency while maintaining high relevance in search results.

Features

  • 4 Search Methods: Basic, Rewrite, Step-back, and Sub-queries retrieval strategies
  • Multi-Agent Support: Configure single or multiple DO Agent instances
  • Chunk-Focused: Returns document chunks with minimal AI overhead for cost optimization
  • Cloudflare Workers Deployment: Scalable serverless deployment
  • Dual-Layer Rate Limiting: Per-IP and global rate limits for cost protection

Usages

usage with local development

python
mcp.run(transport='stdio')  # Tools defined via @mcp.tool() decorator

usage with vscode

{
  "mcpServers": {
    "digital-ocean-agent": {
      "command": "npm",
      "args": ["run", "dev"]
    }
  }
}

usage with cloudflare workers

bash
# 1. Edit wrangler.toml with your endpoint (public)
# 2. Set token as secret (never committed):
wrangler secret put DO_AGENT_TOKEN
# 3. Deploy securely:
npm run deploy

Tools

  1. do_agent_basic_search

    Basic search without query enhancement

  2. do_agent_rewrite_search

    Search with AI query rewriting

  3. do_agent_step_back_search

    Step-back strategy for broader context

  4. do_agent_sub_queries_search

    Sub-queries method for comprehensive retrieval

  5. list_do_agents

    List all configured agent instances

  6. get_current_do_agent

    Get default agent information