> Source: https://wexa.ai/docs/index

# Wexa documentation

Wexa is the enterprise AI context platform. It gives AI agents governed access to everything an
organization knows — its context, knowledge, data and code — through four surfaces that expose the
same capability behind different syntax.

The pages below are grouped by what you are trying to do rather than by which part of the platform
they belong to, because the part of the platform is rarely the thing you know first.

## I want to make a call today

Pick the surface your code already lives in. Each quickstart takes you from nothing to one
successful, verifiable call, and each one covers getting a credential before it asks you to
authenticate anything.

**Quickstart — Wexa MCP server**
Connect an editor or a desktop assistant and watch it call a Wexa tool. No code.

**Quickstart — REST API**
One `curl` command, one credential, one response. Nothing to install.

**Quickstart — TypeScript SDK**
`npm install`, construct a client, make a typed call and handle a typed error.

**Quickstart — Python SDK**
`pip install`, construct a client, make a call and see what a refusal looks like.

## I want to understand it first

Two short pages are worth reading before any quickstart, and one of them will save you the most
common mistake a newcomer makes.

**Overview — What is Wexa**
The four-surface model as a single idea, and the four stores a project holds.

**Before you start — Choosing a surface**
Which of the four fits your stack, and what does not change when you switch.

**Before you start — Scope and server-bound arguments**
Why you never pass your own project, and what happens when you try.

**Concepts — Where your data goes**
The four stores a project holds, told apart. Read this before any one of them.

## I am looking for a specific tool

The tool reference is written by hand and checked against the gateway's own tool registration, so
its arguments match what the platform actually takes. Every tool page carries all four call syntaxes
side by side, and shows a real captured response rather than a sketch.

**Reference — All tools**
Every governed tool, with its arguments and whether your deployment has it.

## I want to know what changed

**Release notes — Changelog**
The current SDK versions and what moved in them.

Not every deployment carries every tool. A tool that is absent is absent on all four surfaces at
once, never on some of them, so a call refused as unknown points at the deployment rather than at
your code. `GET /v1/connection-info` lists what yours has.