Cortex

Cortex is an MCP server by cdeust. Cross-platform persistent memory MCP for Codex, Gemini CLI, Claude Code, and other local MCP hosts. 36 cited neuroscience mechanisms, local-first SQLite/PostgreSQL, hybrid retrieval, decay-based consolidation, and reproducible benchmarks. Claude adds optional automatic lifecycle hooks. It has 75 GitHub stars and ranks #426 of 1287 in Memory & Knowledge by stars.

Category: Memory & Knowledge
Author: cdeust
GitHub stars: 75
This week: Tracking since 18 Oct 2026
Language: Python
License: MIT (open source)
Last update: 2026-10-23 (yesterday)
Links: Source Website PyPI

agent-memory-system
agent-skills
causal-inference
claude-code
codex
cognitive-architecture
cognitive-science
episodic-memory
gemini-cli
hopfield-network
llm-memory
long-term-memory
+8 more

How to install Cortex

Cortex is published on PyPI as hypermnesia-mcp, and it is listed in the official MCP Registry. These steps assume it runs as a local stdio server, which is the usual setup.

Check the README. Confirm the start command and note any API keys or environment variables it needs (pass them with -e KEY=value or an env block). You need uv for uvx.

Add it to Claude Code from a terminal:

claude mcp add Cortex -- uvx hypermnesia-mcp

Or add it to Cursor or Claude Desktop in .cursor/mcp.json or claude_desktop_config.json:

{
  "mcpServers": {
    "Cortex": {
      "command": "uvx",
      "args": [
        "hypermnesia-mcp"
      ]
    }
  }
}

Restart the client and check it. Run claude mcp list, or open the client's MCP settings, and confirm Cortex and its tools are listed. The install guide covers VS Code and Codex.

Also listed in

Maintain this project? Get it featured ($99/mo, pinned at the top of Memory & Knowledge) or see what featuring includes.

Report a problem with this listing

Frequently asked questions

What is Cortex?

Cortex is an MCP server by cdeust in the Memory & Knowledge category. Cross-platform persistent memory MCP for Codex, Gemini CLI, Claude Code, and other local MCP hosts. 36 cited neuroscience mechanisms, local-first SQLite/PostgreSQL, hybrid retrieval, decay-based consolidation, and reproducible benchmarks. Claude adds optional automatic lifecycle hooks. It has 75 GitHub stars and ranks #426 of 1287 in Memory & Knowledge by stars.

Is Cortex free and open source?

Yes. Cortex is published under the MIT license, so you can run, modify and self-host it at no cost. If it connects to a paid service, that service's own pricing and API terms still apply. Check the repository for the exact license text before using it commercially.

What are the alternatives to Cortex?

Other Memory & Knowledge projects on MCP Harbor include open-webui, graphify, claude-mem. Each alternative links to its own page with stars, license and recent activity. The Memory & Knowledge category ranks all 1287 by stars, and you can re-sort it by growth this week.

Alternatives to Cortex

open-webuiTrending★ 156,612

open-webui · Python+650 stars this week (+0.4%)

User-friendly AI Interface (Supports Ollama, OpenAI API, ...)

graphifyTrending★ 134,032

Graphify-Labs · Python+6,459 stars this week (+5.1%)

Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing, every edge explained, no vector store.

claude-memTrending★ 102,636

thedotmack · TypeScript+1,292 stars this week (+1.3%)

Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More

ragflowTrending★ 92,783

infiniflow · Go+413 stars this week (+0.4%)

RAGFlow is a leading open-source Retrieval-Augmented Generation (RAG) engine that fuses cutting-edge RAG with Agent capabilities to create a superior context layer for LLMs

lobehub★ 83,457

lobehub · TypeScript+167 stars this week (+0.2%)

🤯 LobeHub is your Chief Agent Operator, organizing your agents into 7×24 operations by hiring, scheduling, and reporting on your entire AI team.

headroomTrending★ 77,678

headroomlabs-ai · Python+2,109 stars this week (+2.8%)

Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.

All 1287 Memory & Knowledge MCP servers →

Data last updated 25 October 2026. Stars and downloads are refreshed every day from GitHub and npm.