PolyNeural MCP Bundle

You can run the PolyNeural MCP Bundle as an MCP server that bridges your local environment to the PolyNeural.ai backend. It provides persistent memory and knowledge-graph operations through a standard MCP interface, so you can perform entity creation, searches, observations, and more against the PolyNeural.ai platform from your MCP client.

How to use

Install and start the local MCPB server, then connect your MCP client to it. Use the stdio bridge to translate your local MCP messages into HTTP requests against the PolyNeural.ai backend, with authentication forwarded via the API key in headers. You can perform core operations like creating entities and relations, reading graphs, and searching nodes, as well as advanced operations such as getting recent changes or trending entities. Keep your API key secure and set appropriate timeouts to match your network.

How to install

Prerequisites you need before installation are:

• PolyNeural.ai Backend Running: the backend must be reachable (default: https://polyneural.ai)
• Valid API Key: your PolyNeural.ai API key (format: kg_xxxxxxxx)
• Node.js 16+: required to run the extension.

Option 1: As MCPB Bundle (Recommended)

1. Download the latest polyneural-mcpb.mcpb file from the releases page
2. Install using your MCPB-compatible application:

• Claude Desktop: Double-click the .mcpb file to install
• Other MCPB clients: follow your client’s installation process

3. Configure your PolyNeural.ai API key in the bundle settings UI

Option 2: Manual Installation
1. Install dependencies:
```bash
npm install

2. For Claude Desktop: Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "@polyneural/mcpb": {
      "command": "node",
      "args": ["/path/to/mcpb/server/index.js"],
      "env": {
        "API_URL": "https://polyneural.ai",
        "API_KEY": "kg_your_api_key_here",
        "DEBUG": "false",
        "TIMEOUT": "30"
      }
    }
  }
}

Additional configuration and notes

The MCPB bundle exposes environment variables you configure in the client UI or via the environment. You can tailor the backend URL, API key, debug mode, and request timeout to fit your environment.

Environment variables you will typically set:

• API_URL: PolyNeural.ai backend URL (default https://polyneural.ai)
• API_KEY: Your PolyNeural.ai API key (kg_...) (required)
• DEBUG: Enable debug logging (default false)
• TIMEOUT: Request timeout in seconds (default 30)

Configuration

The MCPB bundle supports these user configuration options. You can set these via the MCPB client UI or environment variables.

Troubleshooting

Common issues and fixes:

• Bundle won't start: verify backend is reachable, check API key format (must start with kg_), and enable debug mode in the bundle configuration.
• No tools available in Claude: restart Claude Desktop after installation, check logs for authentication errors, and verify the bundle was installed correctly.
• Connection timeouts: increase TIMEOUT, verify network connectivity to the backend, and confirm the backend MCP endpoints are accessible.

Testing

Manual testing is done by starting the server with your environment variables and ensuring the MCP client can communicate with it. A typical manual run uses environment variables to point to the backend and enable debugging during development.

Development

To modify this MCPB bundle, adjust the server logic in the bridge layer, update bundle configuration, and manage dependencies with your package manager. Build and pack the bundle again when you are ready to test or deploy.

Available tools

create_entities

Store new knowledge entities into the PolyNeural.ai knowledge graph.

create_relations

Create relationships between entities to define the graph structure.

search_nodes

Search the knowledge graph for nodes matching criteria.

open_nodes

Retrieve specific entities by identifier.

read_graph

Get the complete structure of the knowledge graph.

add_observations

Add details or observations to existing entities.

delete_entities

Remove entities from the graph.

delete_relations

Remove relationships between entities.

delete_observations

Remove specific observations from entities.

get_entities_by_identifiers

Bulk retrieval of entities by identifiers.

get_entity_relationships

Fetch connections for a given entity.

get_entities_by_date_range

Query entities by a time range.

get_recent_changes

List recent activity on the graph.

get_trending_entities

Identify popular entities based on activity.

get_frecency_entities

Rank entities by frequency and recency.

search_with_frecency

Smart search with frecency-based ranking.