MCP Protocol Server on .NET 10 Azure Functions for AI Memory Storage and Management

Central Memory MCP

An MCP protocol memory and knowledge graph server implemented based on .NET 10 Azure Functions, providing persistent project memory storage for AI assistants, supporting workspace isolation and entity relationship management.

Knowledge management and memory Developer tools #Knowledge graph #Project memory #MCP service #.NET backend .C#

rating : 2 points

downloads : 6.4K

update time : 2025-12-04

Open Site

What is Central Memory MCP Server?

Central Memory MCP Server is a server that complies with the Model Context Protocol (MCP) standard, specifically designed for AI assistants to store and manage project-related memories and knowledge. It's like an intelligent 'brain' that helps AI remember entities in the project (such as people, concepts, tasks), the relationships between them, and relevant observation records. The server supports multiple workspaces to ensure data isolation and security between different projects.

How to use Central Memory MCP Server?

Using this server is very simple: First, start the server, and then interact with it through the HTTP tool endpoints. You can create or update entities (such as key concepts in the project), establish relationships between entities (such as 'A depends on B'), and query the entire knowledge graph. All operations require specifying the workspace name to ensure organized data.

Applicable scenarios

This server is particularly suitable for AI assistant application scenarios that require long-term memory and context management, such as: • Project management assistant: Remember tasks, personnel, and deadlines in the project. • Research assistant: Track the relationships between research topics, papers, and concepts. • Code development assistant: Understand modules, functions, and dependencies in the codebase. • Content creation assistant: Manage characters, plot lines, and story elements.

Main features

Entity management

Create, update, and store various entities in the project, such as concepts, tasks, people, etc. The system automatically assigns a unique identifier to each entity and supports finding existing entities by name.

Relationship graph

Establish various relationships between entities (such as dependency, inclusion, association, etc.) to build a rich knowledge network. You can query all associated relationships of a specific entity.

Workspace isolation

Supports multiple independent workspaces to ensure complete data isolation between different projects. Each workspace has its own set of entities and relationships.

Observation records

Add observation records (such as status changes, important events, etc.) to entities to help AI understand the history and development process of entities.

Health monitoring

Provides a health check endpoint (/api/health) and a readiness check endpoint (/api/ready) to facilitate monitoring the server status and integrating into the deployment process.

MCP protocol compatibility

Fully complies with the Model Context Protocol standard and can be seamlessly integrated into AI assistants and development tools that support MCP.

Advantages

Persistent storage: All memory data is persistently saved and will not be lost due to the restart of the AI assistant.

Structured knowledge: Organize unstructured information into a structured graph of entities and relationships.

Easy to integrate: Provides services through simple HTTP endpoints without complex configuration.

High performance: Built on Azure Functions, with good scalability and response speed.

Open source and free: Licensed under the MIT license, allowing free use and modification.

Limitations

Alpha stage: Currently in the early development stage, and the functions are still being continuously improved.

Limited search function: The current version lacks advanced search and filtering functions.

Batch operations: Bulk import or export operations are not supported yet.

Semantic understanding: Lacks semantic search ability based on vectors.

Management interface: Currently only has API interfaces, no graphical management interface.

How to use

Environment preparation

Ensure that your system has installed the .NET 10 SDK and Azure Functions Core Tools. If not, you can download and install them from the official Microsoft website.

Get the code

Clone or download the source code of Central Memory MCP Server to a local directory.

Install dependencies

Restore project dependencies and build the solution.

Start the server

Run the server locally on the default port 7071.

Verify the running status

Confirm that the server is running normally through the health check endpoint.

Start using

Now you can call various MCP tools through HTTP requests to manage your knowledge graph.

Usage examples

Project management assistant

In a software development project, use Central Memory Server to track the relationships between features, tasks, and developers.

Research topic tracking

Academic researchers use the server to track the relationships between research topics, related papers, and key concepts.

Novel writing assistant

Writers use the server to manage the relationships between characters, locations, and plot lines in the novel.

Frequently Asked Questions

What is MCP (Model Context Protocol)?

Do I need programming knowledge to use this server?

Where is the data stored? Is it secure?

Can I run multiple workspaces simultaneously?

How can I back up my knowledge graph data?

What's the difference between this server and a vector database?

Related resources

GitHub repository

Get the latest source code, report issues, and contribute.

Model Context Protocol official documentation

Understand the detailed specifications and standards of the MCP protocol.

Azure Functions documentation

Learn how to use and configure Azure Functions.

.NET 10 download

Download and install the .NET 10 runtime and SDK.

Archestra MCP catalog

View the quality and trust scores of this server in the MCP catalog.

🚀 Central Memory MCP Server (.NET 10 Azure Functions)

This is a Model Context Protocol (MCP) compliant memory & knowledge graph server implemented in .NET 10 (Azure Functions isolated worker). It offers durable project memory (entities, relations, observations) for AI assistants with workspace isolation and simple HTTP tool endpoints.

🚀 Quick Start

dotnet restore
dotnet build
func start --port 7071
curl http://localhost:7071/api/health

To read the graph, invoke the read_graph MCP tool from the client and supply the workspaceName parameter.

✨ Features

Implemented MCP Tools

Current function set (alpha stage):

read_graph - returns all entities for a workspace (relations currently queried separately and joined)
upsert_entity - create or update entity (preserves existing Id if name exists)
upsert_relation - create or update relation between two entities (requires GUIDs or resolvable names)
get_entity_relations - list relations originating from a specific entity
Health & Ready endpoints (/api/health, /api/ready)

Planned (not yet implemented): search_entities, search_relations, stats, temporal, batch operations, merge/detect duplicates.

Data Model

Entities table: PartitionKey = WorkspaceName, RowKey = Guid (Id)
Relations table: PartitionKey = WorkspaceName, RowKey = Guid (Id)
Workspaces table (future expansion)

Observations stored as a single delimited string ("||") internally; split into List<string> at read time.

📚 Documentation

Directory Layout

flowchart TD
    A[CentralMemoryMcp.Functions]
    A --> P[Program.cs]
    A --> SR[ServiceRegistration.cs]
    A --> F[Functions]
    F --> GF[GraphFunctions.cs]
    F --> HF[HealthFunctions.cs]
    A --> S[Services]
    S --> KGS[KnowledgeGraphService.cs]
    A --> ST[Storage]
    ST --> TSS[TableStorageService.cs]
    A --> M[Models]
    M --> GM[GraphModels.cs]
    A --> CFG[appsettings.json]
    A --> HOST[host.json]

Usage Notes

Use workspaceName consistently; workspaceId in docs replaced.
Upsert preserves entity identity by lookup on (WorkspaceName + Name).
Relation upsert requires entity GUIDs or resolves names; fails if names missing.