home / mcp / advanced reasoning mcp server
Allows AI to perform advanced reasoning, store the reasoning automatically, and can create system prompts or domain knowledge as libraries.
Configuration
View docs{
"mcpServers": {
"angrysky56-advanced-reasoning-mcp": {
"command": "node",
"args": [
"/path-to/advanced-reasoning-mcp/build/index.js"
]
}
}
}You run a self-contained MCP server that enhances sequential thinking with advanced cognitive features, including meta-reasoning, hypothesis testing, and structured memory. It provides dual storage for reasoning sessions and reusable workflows, making complex problem solving clearer, auditable, and repeatable while you work with an MCP client to drive automated reasoning tasks.
You connect to the local MCP server from your MCP client and begin invoking tools to perform advanced reasoning tasks, memory operations, and structured data management. Use the client to call tools by name, pass your thoughts and hypotheses, and retrieve results that feed back into your ongoing reasoning process.
To integrate with the server, you configure the MCP client to launch or connect to the local process and expose the available tools as endpoints you can call from your code. The server supports a cohesive workflow where you can create memory libraries, reason with meta-cognition, form hypotheses, test them, and store workflows and domain data for reuse.
MCP tools expose capabilities for memory management, reasoning, and structured data storage. You can chain calls to build complex reasoning workflows, retrieve related memories, validate hypotheses with test plans, and store and search workflows or domain data as needed.
MCP client integration uses a local stdio server instance. The server can be started locally and your client can invoke the server through a node-based runtime, passing the appropriate command and arguments to launch the server process.
Example integration focuses on starting a local MCP server process and then referencing it by the assigned server name in your client configuration. This approach keeps the reasoning workflow self-contained and reproducible across sessions.
{
"mcpServers": {
"advanced-reasoning": {
"command": "node",
"args": ["/path-to/advanced-reasoning-mcp/build/index.js"]
}
}
}- Create named memory libraries for project contexts to keep reasoning organized. - Use hypothesis testing to formalize and validate working theories with explicit test plans. - Store workflows and domain data in SystemJSON to enable quick retrieval and reuse in future sessions.
Install prerequisites and build the MCP server locally so you can run it with your MCP client.
Enhanced reasoning with cognitive features: sequential thinking (branching, revisions, dynamic thought counts), confidence tracking (0.0-1.0), reasoning quality assessment (low/medium/high), meta-cognitive reflection, hypothesis formulation and testing, evidence tracking, and memory integration.
Search integrated memory to find related insights, discover connections, and build on previous reasoning sessions with context-aware retrieval.
Create named memory libraries for organized knowledge with separate contexts for different projects or domains.
List available memory libraries with metadata such as name, size, and last modified.
Switch between memory libraries while maintaining session state and context awareness.
Retrieve information about the currently active library, including name, statistics, and session details.
Create structured SystemJSON storage for workflows, domain data, and instructions with domain categorization and tags.
Retrieve stored SystemJSON data by name with metadata and timestamps.
Search through SystemJSON content with relevance scoring across multiple fields.
List all SystemJSON files organized by domain and description with metadata overview.