Integrations Overview
Drop Thunderbit into your favorite AI framework, automation platform, or LLMOps tool
Thunderbit speaks plain HTTP/JSON, so it slots into any agent framework or workflow runner. Each page below is a copy-paste recipe — same auth header, same JSON shapes, same endpoints — adapted to the host platform's idioms.
AI / LLM frameworks
LangChain
Document loader for RAG, or @tool for an agent.
LlamaIndex
Reader for ingestion, FunctionTool for agents.
CrewAI
BaseTool wrapper so any crew agent can read URLs.
Vercel AI SDK
TypeScript tool() for streamText/generateText agents.
Mastra
createTool for Mastra agents and workflows.
Haystack
Custom component for indexing pipelines.
AutoGen
Function tool for AssistantAgent / GroupChat.
MCP
Expose Thunderbit as an MCP server for Claude / Cursor / Windsurf.
Automation / workflow platforms
n8n
HTTP Request node + Wait-for-webhook for batch jobs.
Zapier
Webhooks by Zapier — POST /distill, catch the callback.
Make.com
HTTP module + Webhook trigger, no code.
Pipedream
Code step (Node.js or Python) + HTTP source.
LLMOps / visual builders
Dify
Custom tool via OpenAPI schema — drop the spec, done.
Flowise
Custom Tool node wrapping /distill for chains and agents.
Don't see your platform?
Anything that can make an HTTP POST works. Grab the OpenAPI spec or follow the Quickstart — three endpoints (/distill, /extract, /batch/distill) cover 95% of use cases.