Cognitive Tools MCP

Example Use Cases

• 1

  Meta-cognitively select and combine prompting strategies based on problem-specific context rather than using a fixed reasoning pipeline.

• 2

  Automatically evaluate and score multiple modern cognitive techniques (e.g., cache-augmented reasoning, chain-of-thought variants) and apply the highest-scoring combination.

• 3

  A reusable MCP integration (e.g., Claude Desktop) to orchestrate iterative, tool-aware deliberation with guidance on tool usage and re-deliberation timing.

Description

This MCP implements a dynamic deliberation tool that prompts LLMs to evaluate 15 modern cognitive techniques using a 0.00–0.99 scoring system and a ≥1.53 threshold rule, then selects and applies the most appropriate strategies. It uses a 6-stage framework and 2-round processing to enable meta-cognitive selection of reasoning approaches and iterative improvement. The v10.0.2 release fixes a critical variable reference bug and ensures reliable use of the required input parameter across LLM integrations. It's designed for integration with agent platforms (example: Claude Desktop) and for agents that need adaptive strategy selection rather than hardcoded reasoning flows.