Contextify is a unified memory management MCP server that coordinates AI agent memories across multiple tools ( Claude Code, Codex, Cursor, Windsurf, Gemini, and others) via a REST API and a Web UI. The server runs inside a Docker container and exposes a web UI at http://localhost:8420 and an API at http://localhost:8420/api/v1/. Once started, you can manage memories, perform recall and search operations, and configure which AI tools are integrated. The included CLI commands (contextify) allow you to install, start, stop, update, and query the server status, making it straightforward to operate in development or production environments. You can also interact with the MCP directly through the memory operations such as store, recall, search, and manage contexts to support cross-tool memory sharing and deduping across agents.

Prerequisites:

• Docker installed and running on your machine
• Access to the internet to pull the Contextify container image

Install and run:

1. Pull and run Contextify container (Docker):

Run in detached mode with a named container and port mapping

docker run -d --name contextify -p 8420:8420
-v contextify-data:/var/lib/postgresql/data
ghcr.io/atakanatali/contextify:latest

2. Verify the services are up:

• Web UI: http://localhost:8420
• API: http://localhost:8420/api/v1/
• MCP endpoint: http://localhost:8420/mcp
• Health: http://localhost:8420/health

Optional: If you prefer using the CLI provided by Contextify, install and run the install script as shown in the README:

curl -fsSL https://raw.githubusercontent.com/atakanatali/contextify/main/scripts/install-cli.sh | sh contextify install

Tips and common issues:

• Ensure Docker has enough resources (CPU/memory) allocated for PostgreSQL + Ollama + the MCP server to start reliably.
• The data directory is persisted via a Docker volume (contextify-data) to maintain state across restarts.
• If you change tools or configurations, use contextify status and contextify logs to diagnose connectivity and readiness.
• REST API and Web UI rely on port 8420; if you run behind a proxy or in a containerized environment, adjust network/firewall rules accordingly.
• For CLI interactions, you can use the memory operations (store, recall, search, get, delete, promote) to manage project memories across agents.
• If you need to reconfigure tool integrations (Claude Code, Codex, Cursor, Windsurf, Gemini), run the install flow again or manually adjust the MCP configuration endpoints as documented in the UI.