Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.knowledgestack.ai/llms.txt

Use this file to discover all available pages before exploring further.

Knowledge Stack itself is a hosted (or self-hosted) service. What you install locally is one of four clients that talk to it — pick the one that matches your workflow.

kscli

Best for: terminal users, scripts, CI, bulk ingest.

Python SDK — ksapi

Best for: data pipelines, notebooks, ingestion workers.

TypeScript SDK

Best for: Next.js / Node / Bun / browser apps.

MCP server — ks-mcp

Best for: agents — LangGraph, OpenAI Agents, Claude Desktop, Cursor.

​
CLI — kscli

Requires Python 3.12+ and uv.

​
Python SDK — ksapi

Requires Python 3.10+. Generated from openapi.yaml — fully typed, sync + async clients. 
pip install ksapi

from ksapi import KSApi

ks = KSApi(api_key="sk-user-...", base_url="https://api.knowledgestack.ai")
print(ks.users.me())

​
TypeScript SDK — @knowledge-stack/ksapi

Works in Node 18+, Bun, Deno, and modern browsers. Tree-shakeable. 
npm install @knowledge-stack/ksapi

import { KSApi } from "@knowledge-stack/ksapi";

const ks = new KSApi({
  apiKey: process.env.KS_API_KEY,
  baseUrl: "https://api.knowledgestack.ai",
});
console.log(await ks.users.me());

​
MCP server — ks-mcp

Lets any MCP-compatible agent host (Claude Desktop, Cursor, LangGraph, OpenAI Agents) call Knowledge Stack as tools. 
uv tool install ks-mcp
Then register it with your host — see the MCP server guide for claude_desktop_config.json, Cursor, and LangGraph snippets.

​
Verify your install

kscli login --api-key sk-user-...
kscli whoami
If users.me returns your user object, you’re ready for the Quickstart.
Don’t have an API key yet? Book a demo and we’ll provision a sandbox tenant for you.

​
Self-hosting the server

Running Knowledge Stack itself (not just a client) on your own hardware is covered in the Quickstart’s self-hosted setup and the Deployment section.