Connect QANATIX to any AI model — Claude, GPT, Gemini, LangChain, or your own agent.

Integrations

QANATIX works with any AI model through two protocols:

Protocol	Best for	Latency
MCP (Streamable HTTP)	Claude Desktop, Claude Code, Cursor, any MCP client	Native tool calling
REST API	GPT, Gemini, LangChain, custom agents	Standard HTTP

MCP — native AI tool

QANATIX exposes four MCP tools:

Tool	Description
`qanatix_open_search`	Full-text search across a collection
`qanatix_open_aggregate`	Count, group, and compute stats on records
`qanatix_open_lookup`	Get a specific record by ID
`qanatix_open_collections`	List available collections and their schemas

And two MCP resources:

Resource	Description
`record://\{record_id\}`	Read-only access to a single record by UUID
`schema://\{collection\}/\{record_type\}`	Inferred field names and types for a collection (read-only, auto-detected)

When Claude or Cursor sees a question about data, it calls qanatix_search() automatically and gets back your verified results.

Integration	Protocol	Best for
Python SDK	HTTP (httpx)	Data pipelines, scripts, notebooks
Claude (MCP)	MCP	Claude Desktop + Claude Code
OpenAI (REST)	REST	GPT-4o function calling
LangChain	REST	Tool integration
Cursor	MCP	IDE MCP
Custom agent	MCP/REST	Build your own

Install and start querying in three lines:

import qanatix

qx = qanatix.Qanatix("sk_live_...")
results = qx.search("parts_catalog", "stainless M8 bolt")