---
title: "Cortex overview"
canonical: https://workspace.socra.com/docs/cortex
---

# Cortex overview

Cortex stores owner-governed knowledge that your organization wants coding agents to use before they act. Authorized agents retrieve the same Account-owned Modules across tools and sessions; Cortex does not automatically import an agent host's conversation history or personal memory.

## How Cortex works

Knowledge lives in versioned [**Modules**](/docs/cortex/concepts/modules-and-map). A Module can depend on other Modules. When an agent [flashes a Module](/docs/cortex/concepts/flashing-context), Cortex assembles the requested Module and its dependency closure into one context bundle.

Work lives on the same Module graph. An [**Issue**](/docs/cortex/concepts/issues-and-coordination) identifies a gap or task, records the Module it concerns, and preserves its status, dependencies, comments, and history.

## Who does what

You own and govern the knowledge in your Account. Your coding agents consume that knowledge through the Cortex integration.

The agent host still controls instruction precedence, tool approval, repository access, and command execution. Cortex provides knowledge and coordination data. Cortex does not inspect a repository unless an agent sends repository information through a [Cortex API operation](/docs/cortex/reference/api).

## Use your agents as you already do

Install the Cortex plugin for your coding agent, then keep using it as you already do, including simply chatting. You do not need to create a knowledge base or maintain Modules by hand.

The plugin teaches the agent when and how to use Cortex. The agent flashes existing knowledge when it is relevant to your conversation or work. When either reveals durable knowledge worth keeping, the agent searches the graph and drafts the exact Module change. It does the same when existing knowledge becomes outdated or incomplete.

When conversation or work reveals useful knowledge, the agent updates the relevant Modules or creates missing ones, preserves the reasons, and flashes to verify. It tells you what changed and why. Future authorized agents can flash that knowledge when it becomes relevant.

Cortex does not start work or change Modules in the background. It stays quiet when the conversation or work needs no Account-specific knowledge and reveals nothing worth keeping.

## Choose a starting point

- [Install the Cortex plugin](/docs/cortex/integrations/coding-agents) for your coding agent and keep using it as usual.
- Follow the [Quickstart](/docs/cortex/get-started/first-workflow) when you want to create and flash one Module directly.
- Review [Ways to use Cortex](/docs/cortex/ways-to-use) to compare the web app, Socra CLI, API, and agent integrations.
- [Learn about Modules and the knowledge map](/docs/cortex/concepts/modules-and-map) before designing a larger knowledge structure.