Smart-Thinking
Smart-Thinking
Help your AI check its work.
Smart-Thinking connects your usual AI assistant to tools for checking calculations, finding supporting passages and testing code. Ask in ordinary language; your assistant prepares the check and explains the result.
This package is the small MCP client. MCP is the protocol assistants use to connect to tools. The checking engine runs on the hosted service. You do not need to install a database, run the engine or supply a personal token for public access.
What can you do?
Check a number. Recalculate a discount, scale a recipe or compare complete quotes using exact arithmetic.
Check a source. Find the passage behind a statement and inspect its date, context and origin.
Check a code change. Run approved tests in an isolated environment and see which files were actually tested. A later change needs a new check.
Related MCP server: Tyra Advanced Memory MCP Server
Connect your assistant
You need Node.js 22.16 or newer and Internet access. For an assistant that accepts a mcpServers configuration, add:
{
"mcpServers": {
"smart-thinking": {
"command": "npx",
"args": ["-y", "smart-thinking-mcp@16.3.1"]
}
}
}Save the configuration and reload the MCP connection in your assistant. VS Code uses a different format. Follow the application-specific connection guides for Codex, Claude, Cursor, VS Code or direct HTTP access.
To diagnose a connection from your terminal:
npx -y smart-thinking-mcp@16.3.1 --doctor --allow-networkThe diagnostic checks connectivity and tool discovery. Provider availability is reported separately.
Try these three requests
“Use Smart-Thinking to check €80 minus 20%, then add a 15% fee on the discounted price. Explain the steps.” Expected total: €73.60.
“Use Smart-Thinking to scale 250 g of flour for four people to ten people.” Expected quantity: 625 g.
“Use Smart-Thinking to order these steps: draft before review, review before publish.” Expected order: draft → review → publish.
See eighteen explained examples for sources and code, too.
Data and limits
Requests go to the hosted service. Search and semantic checks can use external providers; submitted code files run in a separate sandbox. The client does not scan your disk. Dossiers and results are stored server-side. Keep anonymous dossier identifiers private: they include access capabilities.
Public access has shared quotas. An exact calculation checks the supplied expression; passing tests cover the executed suite. Model reviews remain fallible. Code review has an internal qualification for bounded, explicit requirements. It can cover the declared semantic review when its evidence and qualification remain valid. This is not external accreditation or a correctness guarantee.
Read the data and trust guide, documentation and evaluations. The website is available in French and English.
MIT license. Source and issues.
Available Tools
1 toolsmartthinkingDInspect
Inputs: thought (required), connections, containsCalculations, generateVisualization, +9 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 1 tool
With only one tool, there is no possibility of confusion or overlap between tools, as there are no other tools to compare it against. The single tool 'smartthinking' stands alone without ambiguity.
Since there is only one tool, it inherently follows a consistent naming pattern with itself. There are no other tools to create inconsistency, making the naming perfectly consistent by default.
A single tool is generally too few for most server purposes, as it limits functionality and scope. Without a description, it's unclear if this minimal set is justified, but typically one tool feels thin and inadequate for meaningful operations.
With no description provided, it's impossible to infer the domain or assess coverage. A single tool with no context suggests severe incompleteness, as there are likely many gaps in functionality for any reasonable server purpose.
Maintenance
Related MCP Connectors
MCP server for building and testing AI agents with multi-model experimentation and insights.
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
MCP server that lets your AI assistant build and deploy apps, hosting each one at a live URL.
An MCP memory server. One memory your agents share — across models, devices and apps.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceAn MCP server that leverages graph structures to perform sophisticated scientific reasoning through an 8-stage processing pipeline, enabling AI systems to handle complex scientific queries with dynamic confidence scoring.3-
- -licenseNot gradedqualityNot gradedmaintenanceA sophisticated MCP server providing advanced memory capabilities with RAG, hallucination detection, and enterprise-grade AI infrastructure for intelligent agent ecosystems.-
- AlicenseAqualityDmaintenanceAn MCP server that enhances sequential thinking with meta-cognitive capabilities including confidence tracking, hypothesis testing, and organized memory storage through graph-based libraries and structured JSON documents.1013MIT
- AlicenseAqualityDmaintenanceAdvanced cognitive thinking MCP server with DAG-based thought graph, multiple reasoning strategies, metacognition, and self-evaluation. A significant evolution beyond sequential-thinking MCP, providing structured deep reasoning with graph-based thought management.627 npm15MIT
Appeared in Searches
- Improved thinking frameworks and methodologies beyond sequential thinking
- A server for finding content related to thinking or contemplation
- A server for finding information about smart technology and intelligence
- OpenRouter thinking or reasoning capabilities
- Using Gemini Pro for Extended Thinking and Reasoning Tasks