Contextual indexes a codebase locally and exposes it through an MCP server and CLI — giving AI coding agents structural call resolution, class-hierarchy awareness, and temporal, git-history-aware retrieval. Grounded answers about what a codebase actually does, not what a model assumes it does.

This repository is the public home of Contextual: documentation, and the place to report bugs or request features. It is proprietary, closed-source software — this repository does not contain, and will never contain, the engine's source code.

Install

uv tool install contextual-engine

pipx works the same way if you don't use uv:

pipx install contextual-engine

Then, from the root of the repository you want Contextual to understand:

contextual init
contextual index

Connect an AI client (Claude Code, Claude Desktop, Cursor, Windsurf, and others), and it can now ground its answers in your actual codebase — not a guess. Full walkthrough: Getting Started.

What is Contextual?

Contextual is a local-first context engine for AI coding agents: a CLI and MCP server that build a structural + semantic + temporal understanding of a codebase on-device, then serve it to whatever agent is asking. No code ever leaves the machine.

Use it for onboarding onto an unfamiliar codebase, keeping an AI agent grounded in a large monorepo, code review context, or any workflow where "the model guessed" isn't good enough.

%%{init: {'theme':'neutral'}}%%
flowchart LR
    A[Your repository] --> B["contextual index<br/>local, on-device"]
    B --> C[MCP server]
    C --> D["AI client<br/>Claude Code · Cursor · Windsurf · ..."]

Documentation

Full docs: contextuallabs.dev/docs — or browse them directly in this repo:

• docs/cli/ — the Python CLI

• docs/mcp-server/ — the MCP server and its tools

• docs/web-dashboard/ — the web dashboard

• docs/concepts/ — architecture and cross-cutting concepts

• docs/account/ — licensing, trial, device management

• docs/changelog/ — release notes

Report a bug or request a feature

GitHub Issues is the official channel for bug reports, docs feedback, and feature requests. Contextual is closed-source, proprietary software, so this isn't a place for code contributions — it's where the community side of this project actually lives: tell us what's broken, what's missing, or what you'd like to see.

Open an issue and pick the template that matches. Typical first response is 1–3 business days — see SUPPORT.md for the full detail, and SECURITY.md to report a vulnerability privately instead.

For anything else: team@contextuallabs.dev

License

All rights reserved — this repository's content is publicly viewable, not publicly licensed. No commercial use, copying, or redistribution is permitted. See LICENSE. Contextual (the CLI/MCP server) is closed-source, licensed software — see contextuallabs.dev for terms.