Mcp Thinkgate
Routes Claude prompts through automatic complexity classification before they hit the model. Uses a fast Haiku call to bin incoming messages into fast, think, or ultrathink tiers, then returns the appropriate effort level and model recommendation. Exposes a classify_prompt tool when running as an MCP server in Claude Desktop, or works as a library import in agent frameworks. The actual value is in the classifier prompt itself, which decides the boundary between what needs extended thinking and what doesn't. Ships with a rule-based fallback if you skip the API key. Built for agent developers running mixed workloads where you're wasting tokens on trivial requests or under-allocating on complex ones.
- GitHub stars
- 2
- tool
- 1
- auth
- Secret needed
- transport
- STDIO
- env var
- 1
- last updated
- Feb 19, 2026
Tools
1 toolsPublic tool metadata for what this MCP can expose to an agent.
classifyPrompt
No parameter schema in public metadata yet.
Sponsors
Agent CaffeineSponsoredClose your lid, come back to finished work
SnugSponsoredThe clipboard manager for the AI eraPromote your product hereReach 125k+ visitors monthly
WaxSponsoredBlock anything with supervision, without erasing your iPhonePromote your product hereReach 125k+ visitors monthly
knobsSponsoredSee your site on any phone, any day, in any languageConfiguration
ANTHROPIC_API_KEYsecretOptional. Enables AI-powered classification via Haiku (more accurate). Falls back to rule-based classification if not provided. Get a key at https://console.anthropic.com
Readme
ThinkGate
Automatic reasoning mode selection for Claude agents.
The problem
You built an AI agent. It handles everything — status checks, quick lookups, complex architecture questions, deep debugging sessions. But under the hood it runs every single message through the same model with the same thinking settings.
That means you're burning extended thinking tokens on "what time is it in Tokyo?" and getting shallow answers on "help me design the entire auth system."
You could manually tag requests — ULTRATHINK: before the hard ones. But you forget. Your users definitely won't do it. And if you're building agents for other people, you can't train every end user to manage thinking modes.
ThinkGate fixes this at the infrastructure layer. It sits between the incoming message and your model call, classifies the complexity in ~200ms, and returns exactly which model and thinking depth to use. Automatically. Every time.
Who this is for
Details
- Repository
- github.com/tjp2021/mcp-thinkgate
- Category
- AI & LLM Tools
- Package
mcp-thinkgate- Transport
- STDIO
- Auth
- Required
- Registry
- MCP registry, active
Related AI & LLM Tools MCP servers
View all
Pseudonym Mcpwoladi/pseudonym-mcpLocally pseudonymizes sensitive text before cloud LLM work and restores the tokens afterward.2stars
Pop Pay100xpercent/pop-payRuntime security for AI agent commerce. CLI + MCP server blocks hallucinated purchases.1stars
Forgevoxellinc/forge-mcpMCP server for Forge, Voxell's embedding API. Voxell's Ingot-8B-R3 ranks #1 for English on MTEB.1stars
Amberambermem/amberLong-term memory for AI assistants: permanent, portable across clients, searchable and exportable.1stars
Queryshieldbch1212/queryshieldSecure SQL proxy for AI agents — NL→SQL, AST safety, per-agent RLS, audit log.1stars