kobolingfengkobolingfengCommunity·1 Agent Skills Included

ntk

Adaptive task routing that cuts token costs by 90%

Routes coding, analysis, and design tasks through adaptive pipelines matched to their real complexity. Compresses context between steps so each stage only receives the minimum information it needs. Eliminates wasted tokens on simple tasks by running cheap models for 95% of work and reserving strong models for planning. Connects to VS Code, Claude Desktop, and OpenClaw via MCP for instant use.
npx skills add kobolingfeng/ntk --all -g -y

All Skills in This Repository (1)

Pure Emerald Level Indicators

Frequently Asked Questions

FAQPage Schema
How to install NTK?

Run `npx skills add kobolingfeng/ntk --all -g -y` in your terminal to install it globally, then add your own OpenAI-compatible API key in the configuration.

How does NTK reduce token costs?

It classifies each task's complexity first, sends simple tasks straight to execution, and runs over 95% of tokens on cheap models, cutting costs by up to 90% versus using strong models for everything.

What tasks can NTK handle?

It handles code generation, bug fixing, refactoring, translation, API design, log analysis, and system architecture design, automatically picking the right pipeline depth for each.

Does NTK work with Claude Desktop and VS Code?

Yes. NTK runs as a native MCP server, so it plugs directly into Claude Desktop, VS Code Copilot, OpenClaw, and any other MCP-compatible client.

Do I need my own API key to use NTK?

Yes. NTK does not bundle any keys, so you configure your own OpenAI-compatible endpoint, including local options like Ollama or LM Studio.

Related Repositories in Software Engineering

View All in Software Engineering