Cortex FAQs

Question 1

How does Cortex capture and manage project memory?

Accepted Answer

Cortex silently monitors your AI coding sessions in real-time, building a 3-layer memory architecture: Working (hot), Episodic (warm), and Semantic (cold). It auto-generates Architectural Decision Records (ADRs) and injects the most relevant context into your AI assistant using native integrations like CLAUDE.md or MCP.

Question 2

Does Cortex require an API key to function?

Accepted Answer

No, Cortex works out-of-the-box with basic local pattern-matching extraction. An optional, free API key (e.g., Gemini) can be added for deeper, LLM-powered memory extraction, enhancing the intelligence and accuracy of the captured context.

Question 3

Is my data private and secure with Cortex?

Accepted Answer

Yes, absolutely. Cortex is designed with privacy in mind: all data is stored 100% locally in a '.cortex/' folder on your machine. There is no telemetry, no cloud storage (beyond your chosen LLM API for advanced extraction), and you retain full control over your data.

Question 4

Which AI coding assistants are supported by Cortex?

Accepted Answer

Cortex integrates natively with Claude Code via CLAUDE.md auto-injection and session watching. It also provides MCP server support for Cursor, Cline, Zed, and Continue. For Copilot, it passively enhances its context if CLAUDE.md is present in your project.

Question 5

What is Cortex and how does it help developers?

Accepted Answer

Cortex gives AI coding assistants a persistent memory, eliminating the need to re-explain project context, architecture, past decisions, or known bugs in every session. It automatically captures this information and injects it into your next AI interaction, saving 15-30 minutes per session.

Cortex

Cortex

主な機能

ユースケース

主な機能

ユースケース