MaxMemCp: The Living Memory MCP Server

MaxMemCp bridges the gap between static codebases and dynamic LLM agents by creating a centralized, queryable intelligence layer. It stores Intent (Design Docs) alongside Implementation (Code) and prevents "Context Amnesia".

🚀 Quick Start

1. Installation

We recommend using uv.

uv pip install -e .

2. Configuration

Set your LLM Provider.

=== "Bash"

```bash
export MAXMEM_LLM_PROVIDER_API_KEY="sk-..."
# OR 
export MAXMEM_LLM_PROVIDER="ollama"
```

=== "PowerShell"

```powershell
$env:MAXMEM_LLM_PROVIDER_API_KEY="sk-..."
# OR
$env:MAXMEM_LLM_PROVIDER="ollama"
```

3. Run Server

python main.py

📚 Documentation

• Features & Capabilities: Learn about the "Surveyor", "Living Memory", and Hybrid Retrieval.
• Architecture: High-level design, Tech Stack, and Layers.
• Setup Guide: Detailed installation and environment configuration.
• Usage Guide: How to use the MCP tools (initialize_memory, query_memory).
• Technical Spec: Data schemas and internal logic.

🛠 Features at a Glance

• Surveyor Workflow: Auto-generates a Project Manifest.
• Hybrid Retrieval: Combines Vector Search with Grep Fallback for 100% precision.
• Local Support: Full support for Ollama models.
• Context-Aware: Queries are grounded in the project architecture.