Apache AGE MCP

We need to translate the given text from English to Turkish. The text describes an MCP server for Apache AGE graph databases. We must preserve product names, protocol names, URLs, numbers, and technical terms. The name "Apache AGE MCP" appears in the text, so we keep it as is. Also preserve "MCP", "CRUD", "Cypher", "vis.js", "Graph RAG", "F#/.NET", "dotnet tool", "BenchmarkDotNet", "NuGet", "MIT license", "neftedollar.com/age-mcp". Translate the rest naturally. The text inside <text> is: "MCP server for Apache AGE graph databases on PostgreSQL. **21 MCP tools** — the most comprehensive Apache AGE MCP server (graph CRUD, Cypher queries, batch transactions, semantic search, Graph RAG, vis.js visualization, export/import) - **F#/.NET** — the only non-Python Apache AGE MCP server, installs as a single dotnet tool - **

GitHub

Dokümantasyon

age-mcp

MCP server for Apache AGE graph databases. Lets AI assistants (Claude, ChatGPT, Copilot, etc.) query and mutate graph data via the Model Context Protocol.

Built with F# on .NET 10 -- 1,300 lines, 21 tools, zero Python dependencies.

Quick Start

# 1. Start the database
docker compose up -d

# 2. Install the tool
dotnet tool install --global AgeMcp

# 3. Run
AGE_CONNECTION_STRING="Host=localhost;Port=5435;Database=agemcp;Username=agemcp;Password=agemcp" age-mcp

Claude Desktop / Claude Code

{
  "mcpServers": {
    "age-mcp": {
      "type": "stdio",
      "command": "age-mcp",
      "env": {
        "AGE_CONNECTION_STRING": "Host=localhost;Port=5435;Database=agemcp;Username=agemcp;Password=agemcp",
        "TENANT_ID": "default"
      }
    }
  }
}

Tools (21)

Graph Management

Tool	Description
`get_or_create_graph`	Get or create a graph by name
`list_graphs`	List all graphs (tenant-scoped)
`drop_graphs`	Drop one or more graphs

Vertices & Edges

Tool	Description
`upsert_vertex`	Insert or update a vertex (merge on ident)
`upsert_edge`	Insert or update a directed edge
`upsert_graph`	Batch upsert vertices + edges (transactional)
`drop_vertex`	Remove a vertex and all its edges
`drop_edge`	Remove an edge by ident

Query

Tool	Description
`cypher_query`	Execute a read Cypher query
`cypher_write`	Execute a write Cypher query, returns affected count
`search_vertices`	Search by label and/or property
`search_edges`	Search edges by label
`get_neighbors`	N-hop traversal (1-5 hops, directional)
`get_schema`	All node labels and counts

Export / Import

Tool	Description
`export_graph`	Export graph as JSON
`import_graph`	Import from JSON (creates graph if needed)

Visualization & Search

Tool	Description
`generate_visualization`	Interactive vis.js HTML graph
`semantic_search`	Vector similarity search (pgvector)
`graph_context`	Graph RAG: semantic seeds + N-hop expansion

OpenBrain Bridge

Tool	Description
`sync_to_openbrain`	Export vertices as OpenBrain memories
`import_from_openbrain`	Build graph from OpenBrain memories

Configuration

Variable	Required	Default	Description
`AGE_CONNECTION_STRING`	yes	localhost test DB	Npgsql connection string
`TENANT_ID`	no	`default`	Tenant prefix for graph names
`EMBEDDING_API_URL`	no	--	OpenAI-compatible embedding API
`EMBEDDING_API_KEY`	no	--	API key for embeddings
`EMBEDDING_MODEL`	no	`text-embedding-3-small`	Embedding model name
`EMBEDDING_DIMENSIONS`	no	`384`	Vector dimensions

Docker

The included Docker setup runs PostgreSQL 17 + Apache AGE 1.6.0 + pgvector:

docker compose up -d

Versions and credentials are configurable via .env (see .env.example):

PG_MAJOR=17  AGE_VERSION=1.6.0  DB_PORT=5435  docker compose up -d

Data Compatibility

Designed as a drop-in replacement for agemcp (Python). Same tenant prefix (t_{TENANT_ID}__), same vertex ident property, same edge start_ident/end_ident properties. Existing data works without migration.

Performance

BenchmarkDotNet on Apple M1 Pro, .NET 10.0.5:

Operation	Latency	Allocated
list_graphs	62 ns	216 B
get_schema	117 ns	344 B
cypher_query (1 vertex)	1.0 ms	58 KB
search_vertices	1.0 ms	36 KB
get_neighbors (depth=1)	92 ms	233 KB
export_graph (40 entities)	71 ms	472 KB

Building from Source

git clone https://github.com/Neftedollar/age-mcp.git
cd age-mcp
dotnet build

# Run directly
dotnet run

# Or install as tool
dotnet pack -c Release
dotnet tool install --global AgeMcp

Dependencies (FsMcp.Core, FsMcp.Server, Fyper) are restored from NuGet automatically.

License

MIT