Integrations
Every endpoint is a GET with a header, so there is nothing to install. MCP clients get all11 tools from one config block; frameworks need about ten lines.
{
"mcpServers": {
"unlob": {
"command": "npx",
"args": ["-y", "@unlob/mcp"],
"env": { "UNLOB_API_KEY": "ulb_your_key_here" }
}
}
}MCP clients
One config block gives the agent all eleven tools, including the coverage graph.
Claude Code
Add unlob to Claude Code as an MCP server and give it all eleven tools in one config block.
Claude Desktop
Configure unlob as an MCP server in Claude Desktop for web search and graph traversal in chat.
Cursor
Add unlob to Cursor as an MCP server so the editor agent can search documentation and the open web.
ChatGPT
Connect unlob to ChatGPT through MCP for retrieval you control rather than the built-in web tool.
Agent frameworks
A plain GET with a header — a custom tool is about ten lines, and you keep control of the filters.
LangChain
Wrap the unlob search API as a LangChain tool in a few lines — it is a plain HTTP GET.
LlamaIndex
Use unlob as a retriever in LlamaIndex — metadata first, then fetch the documents you actually selected.
Vercel AI SDK
Define unlob as a tool in the Vercel AI SDK with a Zod schema and a fetch call.
OpenAI Agents SDK
Register unlob as a function tool so your agent has retrieval you control rather than the built-in web tool.
CrewAI
Give CrewAI agents a shared unlob search tool with filters appropriate to each role.
Automation
No-code workflows, using a standard HTTP node.
Not listed?
Everything is a GET with a header, so any HTTP client works. The API reference documents every endpoint and parameter.
