archiledger

Archiledger is an MCP server that provides a persistent memory graph for your AI assistant. It stores MemoryNotes as nodes in a knowledge graph and supports typed links between notes (e.g., LINKED_TO) to build a connected memory with context and provenance. The MCP tools exposed by Archiledger let you create notes, retrieve notes by ID, search by tags, delete notes, and manage graph relationships. You can also perform graph exploration operations such as reading the entire graph, listing linked notes, retrieving all tags, and running semantic similarity searches over the note embeddings. This enables your LLM-based assistants to recall prior conversations, relate ideas across notes, and discover non-obvious connections across your data.

Prerequisites:

• Docker installed and running (recommended for Archiledger as documented) or Java 21+ with Maven if you prefer building from source
• Internet access to pull Docker image or build dependencies

Option A: Run via Docker (recommended and documented in this repo)

1. Ensure Docker is installed and you have access to the registry where the image is hosted.
2. Start Archiledger with: docker run -p 8080:8080 registry.hub.docker.com/thecookiezen/archiledger:latest
3. The MCP server will be available at http://localhost:8080 by default.

Option B: Build from source (advanced)

1. Install Java 21+ and Maven.
2. Clone the repository and navigate to the project root.
3. Build the project: mvn clean package
4. Run the built JAR (example): java -jar mcp/target/archiledger-server-0.0.1-SNAPSHOT.jar
5. Access the MCP endpoints at http://localhost:8080 (default port).

Note: If you plan to use the embedded LadybugDB or external persistence, follow the README guidance for enabling the respective profiles or Docker environment variables.

Tips and caveats:

• This server currently has no authentication; use in trusted environments only.
• When using Docker, you can persist data by mounting a host directory into the container and/or by setting LADYBUGDB_DATA_DIR to a desired path inside the container.
• If you expect larger graphs, consider increasing memory or using the provided data-dir persistence to avoid data loss on container restarts.
• The README mentions a BugScope visualization option when using LadybugDB with Neo4j; enable the appropriate profile and point BugScope to your data directory if you need visualization.
• The embedded embedding generation relies on a model downloaded via Spring AI tooling; ensure network access or pre-downloads as necessary for offline runs.
• If you run into port conflicts, adjust the Docker run port mapping (e.g., -p 8081:8080) and access via the new port.