Audrey
Local-first memory firewall and MCP server for AI agents. Audrey provides recall, memory capsules, preflight checks, reflexes, validation, tool-trace learning, contradiction handling, and SQLite/sqlite-vec storage.
What is Audrey?
Audrey is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to local-first memory firewall and mcp server for ai agents. audrey provides recall, memory capsules, preflight checks, reflexes, validation, tool-trace learning, contradiction handling, and sqlite/sqlit...
Local-first memory firewall and MCP server for AI agents. Audrey provides recall, memory capsules, preflight checks, reflexes, validation, tool-trace learning, contradiction handling, and SQLite/sqlite-vec storage.
This server falls under the Knowledge & Memory category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- Local-first memory firewall and MCP server for AI agents. Au
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx audreyConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use Audrey
Audrey is a local-first memory firewall and MCP server that gives AI agents persistent, structured memory backed by SQLite and sqlite-vec — entirely on your machine with no cloud dependency. It provides semantic recall, preflight safety checks before risky actions, tool-trace learning from failed commands, memory capsules, contradiction handling with decay, and reflexes (rule-based guards). Audrey integrates with Claude, Codex, and other MCP-compatible agents to maintain context and learned behavior across sessions.
Prerequisites
- Node.js 20+ installed
- An MCP-compatible AI client such as Claude Desktop, Claude Code, or Codex
- Optional: AUDREY_EMBEDDING_PROVIDER set to 'gemini' or 'openai' with corresponding API key for cloud embeddings (defaults to local CPU embeddings)
- Optional: Docker if using the containerized deployment path
Run the health check
Before installing, run the doctor command to verify your Node.js version and system readiness.
npx audrey doctorInstall Audrey for your agent host
Use the install command to auto-configure Audrey for your chosen agent host. Use --dry-run first to preview changes.
npx audrey install --host codex --dry-run
npx audrey installConfigure environment variables
Set environment variables to control storage location, agent identity, embedding provider, and authentication. The most important are AUDREY_DATA_DIR and AUDREY_AGENT.
export AUDREY_DATA_DIR=~/.audrey/data
export AUDREY_AGENT=my-agent
export AUDREY_EMBEDDING_PROVIDER=local
export AUDREY_LLM_PROVIDER=anthropic
export AUDREY_PORT=7437Add to Claude Desktop config
Add Audrey to your MCP client configuration. The npx command will auto-download and run the server.
{
"mcpServers": {
"audrey": {
"command": "npx",
"args": ["audrey"],
"env": {
"AUDREY_DATA_DIR": "/Users/you/.audrey/data",
"AUDREY_AGENT": "claude-agent",
"AUDREY_EMBEDDING_PROVIDER": "local"
}
}
}
}Run a demo scenario
Try the built-in demo to see how Audrey learns from repeated failures and applies that knowledge in future preflight checks.
npx audrey demo --scenario repeated-failureUse guard mode for high-risk commands
Wrap any shell command with audrey guard to have Audrey run a memory preflight check and reflexes validation before execution.
npx audrey guard --tool Bash "npm run deploy"Audrey Examples
Client configuration
Claude Desktop configuration for Audrey with local embeddings and a dedicated data directory.
{
"mcpServers": {
"audrey": {
"command": "npx",
"args": ["audrey"],
"env": {
"AUDREY_DATA_DIR": "/Users/you/.audrey/data",
"AUDREY_AGENT": "claude-agent",
"AUDREY_EMBEDDING_PROVIDER": "local",
"AUDREY_PORT": "7437"
}
}
}
}Prompts to try
Example prompts that leverage Audrey's memory, recall, and preflight capabilities.
- "What do you remember about the last time we deployed to production?"
- "Before running the database migration, check your memory for any past issues with this operation"
- "Store this as a lesson: always run tests before merging to main"
- "Recall everything you know about our API rate limits"
- "Consolidate today's session memories into a summary"Troubleshooting Audrey
npx audrey fails with a Node.js version error
Audrey requires Node.js 20 or higher. Run node --version to check, then upgrade via nvm: nvm install 20 && nvm use 20.
Embeddings are slow or failing with local provider
The local embedding provider runs CPU-only inference which can be slow on first use. For faster results set AUDREY_EMBEDDING_PROVIDER=openai or AUDREY_EMBEDDING_PROVIDER=gemini and provide the corresponding API key.
Agent memory is leaking between different projects
Set a distinct AUDREY_DATA_DIR and AUDREY_AGENT value per project or agent identity. Each unique data directory is an isolated memory store.
Frequently Asked Questions about Audrey
What is Audrey?
Audrey is a Model Context Protocol (MCP) server that local-first memory firewall and mcp server for ai agents. audrey provides recall, memory capsules, preflight checks, reflexes, validation, tool-trace learning, contradiction handling, and sqlite/sqlite-vec storage. It connects AI assistants to external tools and data sources through a standardized interface.
How do I install Audrey?
Follow the installation instructions on the Audrey GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with Audrey?
Audrey works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is Audrey free to use?
Yes, Audrey is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.
Audrey Alternatives — Similar Knowledge & Memory Servers
Looking for alternatives to Audrey? Here are other popular knowledge & memory servers you can use with Claude, Cursor, and VS Code.
MemPalace
★ 52.6kA local AI memory system that stores all conversations verbatim and organizes them into navigable structures. It provides 19 MCP tools for AI assistants to search and retrieve past decisions, debugging sessions, and architecture debates automatically
Kratos
★ 25.7k🏛️ Memory System for AI Coding Tools - Never explain your codebase again. MCP server with perfect project isolation, 95.8% context accuracy, and the Four Pillars Framework.
Context Mode
★ 15.4kAn MCP server that preserves LLM context by intercepting large data outputs and returning only concise summaries or relevant sections. It enables efficient sandboxed code execution, file processing, and documentation indexing across multiple programm
Memu
★ 13.7kMemory for 24/7 proactive agents like OpenClaw.
MemOS
★ 9.3kMemOS (Memory Operating System) is a memory management operating system designed for AI applications. Its goal is: to enable your AI system to have long-term memory like a human, not only remembering what users have said but also actively invoking, u
Everos
★ 5.4kBuild, evaluate, and integrate long-term memory for self-evolving agents.
Browse More Knowledge & Memory MCP Servers
Explore all knowledge & memory servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up Audrey in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use Audrey?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.