MCP-Mem0

MCP-Mem0 is an MCP server that integrates Salesforce development capabilities into AI agents through CumulusCI. Developed for Salesforce developers, release engineers, and automated coding assistants, the server bridges large language models with the CumulusCI toolset so agents can execute development operations without manual command line entry. It allows assistants running in compatible environments to create and open scratch orgs across different configurations, inspect active environments, deploy metadata, retrieve remote changes, and execute Apex test suites. Additionally, the server exposes framework documentation as Model Context Protocol resources, providing agents with contextual design patterns, trigger architectures, logging standards, and platform cache management guidelines. It also includes a generic task handler that inspects and runs arbitrary CumulusCI commands with dynamically collected parameters, enabling technical teams to safely delegate routine Salesforce administrative and continuous integration tasks to AI agents.

Category: AI & LLM Tooling

Tags: long-term memory, mem0, personalization, user-context

Visit MCP-Mem0

How to install and configure MCP-Mem0

Follow these steps to install and configure the server using uv or Docker as documented in the project repository. 1. Ensure Python 3.12+ and CumulusCI are installed and configured with your Salesforce Dev Hub org: bash pip install cumulusci cci org connect <your-dev-hub> 2. Clone the repository and install dependencies with uv: bash git clone https://github.com/jorgesolebur/mcp-cci cd sfcore-th-dev uv pip install -e . 3. To run with stdio transport in Claude Desktop, add the server configuration to your claude_desktop_config.json: json { "mcpServers": { "sfcore-th-dev": { "command": "python", "args": ["src/main.py"], "env": { "TRANSPORT": "stdio" } } } } Alternatively, you can run the server via Docker using ghcr.io/jorgesolebur/mcp-sfcore-th-dev:latest with port 8050 mapped for SSE transport.

What you can do with MCP-Mem0

• Create development, feature QA, and beta scratch orgs automatically through standardized CumulusCI flow commands. - Execute local Apex test suites in designated Salesforce environments and report execution results to developers. - Deploy local metadata directly to connected orgs without manually constructing complex CLI commands. - Retrieve metadata changes made in remote scratch orgs into the local project repository. - Access architectural guidance for Apex triggers, logging frameworks, and platform cache management via MCP resources.

Key facts

• Open Source
• https://github.com/jorgesolebur/mcp-cci
• AI & LLM Tooling, AI Memory & Context
• long-term memory, mem0, personalization, user-context

Part of MCP Servers

What is MCP-Mem0?

MCP-Mem0 is an open-source Model Context Protocol server that exposes CumulusCI commands and Salesforce development resources to AI agents. It enables coding assistants to manage scratch orgs, run Apex tests, deploy metadata, retrieve changes, and access Salesforce architectural patterns without requiring developers to manually execute CLI flows.

Which MCP clients work with MCP-Mem0?

The server supports both stdio and Server-Sent Events transports, allowing it to integrate with Anthropic Claude Desktop, Windsurf, n8n, and any other Model Context Protocol client capable of initiating local Python processes or connecting to SSE endpoints.

What tools are available through MCP-Mem0?

The server includes tools for checking CumulusCI installation, creating development, feature, and beta scratch orgs, listing connected orgs, opening orgs in a browser, running Apex tests, deploying metadata, retrieving remote changes, and executing arbitrary generic CumulusCI tasks with automated parameter inspection.

Is MCP-Mem0 open source?

Yes, MCP-Mem0 is an open-source project hosted on GitHub. Developers can review the source code, customize tool implementations, and deploy it locally or via pre-built Docker containers.