mcp-server-systems-thinking
If you are the rightful owner of mcp-server-systems-thinking and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.
The Systems-Thinking MCP server is designed to facilitate AI agents in submitting and iterating on systems-thinking models using a structured JSON format, ensuring comprehensive validation and atomic persistence.
The Systems-Thinking MCP server is a specialized server that enables AI agents to submit a complete systems-thinking representation in the form of a JSON document over HTTP. The server is built to validate the structure of the document, persist the latest version atomically, and provide feedback on validation gaps, allowing the agent to iterate until the document is complete. This approach is inspired by Donella Meadows-style systems analysis, focusing on boundaries, stocks/flows, feedback loops, and leverage points. The server employs a rigid I/O contract to improve reasoning quality, similar to sequential-thinking servers. The architecture leverages Node.js with TypeScript, FastMCP for HTTP streaming transport, and Zod for validation, with data persistence managed through an in-memory cache and nightly flushes to Postgres.
Features
- Validation & Gap Detection: Ensures the JSON document conforms to the required schema and identifies missing fields or inconsistencies.
- Atomic Persistence: Stores the latest version of the document atomically, ensuring data integrity.
- HTTP Streaming Transport: Utilizes FastMCP for efficient data transmission.
- Basic Observability: Includes request logs and a health endpoint for monitoring.
- In-Memory to Postgres Persistence: Uses an in-memory map with nightly flushes to Postgres for data storage.
Tools
systems_thinking_writer
A tool for submitting a full systems-thinking JSON document for validation and persistence.