
Cotforce Mcp
Forces small language models to use step-by-step reasoning instead of guessing. Works through MCP's sampling protocol with Claude Desktop or via direct HTTP calls to local models like Gemma running in LMStudio. The single cot_force tool wraps any prompt in strict chain-of-thought enforcement, parsing structured reasoning and results from JSON responses. Includes adaptive retry logic with temperature ramping, token budget tracking with tiktoken, and five fallback parsers for when models output malformed JSON. Configurable via environment variables for timeout, retry count, and model selection. Ships with model-specific prompts tuned for Claude, GPT-4, Gemini, and Grok. The test suite covers 151 cases across parser pipeline, schema validation, and retry mechanics.
- tool
- 1
- transport
- STDIO
- env vars
- 3
- last updated
- May 7, 2026
Tools
1 toolsPublic tool metadata for what this MCP can expose to an agent.
solve_problemSolves a problem using step-by-step Chain-of-Thought reasoning, taking a prompt string and returning a structured CoT result.
Solves a problem using step-by-step Chain-of-Thought reasoning, taking a prompt string and returning a structured CoT result.
No parameter schema in public metadata yet.
Sponsors
Configuration
MODELLM call mode: auto, direct, or sampling
API_BASE_URLBase URL for direct HTTP mode (e.g. http://localhost:1234/v1)
MODELModel name or hint (e.g. gemma-4-e4b-it-mlx)
Readme
CotForce-MCP
"Give brains to your small models."
CotForce enforces step-by-step Chain-of-Thought, turning 4B parameter models into methodical reasoners.
Why this exists
A 4-billion-parameter Gemma cannot solve SEND + MORE = MONEY. It's a classic cryptarithmetic puzzle — 8 unique digits, 5 columns, 4 carry values. A bare 4B model guesses randomly. It hallucinates digits. It loses track of carries after column 2.
The same model, with CotForce:
Step 1: Analyze the leftmost column. S+M+C3 = MO. Max sum is 19998. ∴ M=1.
Step 2: S+1+C3 = 10+O. With M=1 and carry, O must be 0.
Step 3: D+E = Y+10C1 → C1=1. Now R+C1=9 → C1=0→R=9 (used), C1=1→R=8.
...
Step 11: All digits assigned. 9567 + 1085 = 10652. Verified.
11 structured reasoning steps. Zero hallucinations. Correct answer.
CotForce doesn't make small models smarter. It forces them to think before they speak — which is often all they need.
Details
- Repository
- github.com/islobodan/cotforce-mcp
- Package
@slbdn/cotforce-mcp- Transport
- STDIO
- Registry
- MCP registry, active