Claudish is a CLI proxy tool that allows Claude Code to interface with any model available on OpenRouter, providing access to Grok, Gemini, MiniMax, and more.

Why should I use sub-agents instead of running Claudish directly?

Running Claudish in the main context adds over 10,000 tokens of overhead (including reasoning and tool outputs), which quickly exhausts your context window. Sub-agents keep your main conversation clean.

What models are recommended for coding tasks?

For fast iterations, x-ai/grok-code-fast-1 is highly recommended. For complex architecture and deep reasoning, google/gemini-2.5-flash or openai/gpt-5 are the preferred choices.

How do I install the necessary tools for this skill?

You need the Claudish CLI, which can be installed via npm with 'npm install -g claudish'. You also need an OpenRouter API key set in your environment variables.

Can I use Claudish for UI and frontend design?

Yes, by selecting vision-capable models like Qwen-VL through Claudish, you can perform visual analysis and UI implementation tasks directly within your workflow.

Claudish Usage & Model Proxy

Name: Claudish Usage & Model Proxy
Author: MadAppGang

byMadAppGang

•

177

Herramientas para Desarrolladores

Integrates external OpenRouter models into Claude Code workflows while maintaining context efficiency through intelligent sub-agent delegation.

Acerca de

Claudish Usage is a specialized framework for running Claude Code with OpenRouter's extensive model library—including Grok, GPT-5, and Gemini—without overwhelming the main conversation's context window. It enforces a strict sub-agent execution pattern that keeps main sessions lean, provides a curated model selection guide based on task types (coding, reasoning, vision), and includes implementation patterns for file-based instruction passing. This skill is essential for developers wanting to leverage multi-model workflows, cost-optimize their AI usage, or utilize specific model strengths for specialized software development tasks.

Características Principales

Access to 100+ OpenRouter models including Grok-3, GPT-5, and Gemini 2.0.
Support for both direct agent invocation and file-based instruction workflows.
Intelligent agent selection matrix for specialized tasks like architecture and code review.
Integrated CLI tools for model fuzzy searching, top-model recommendations, and cost tracking.
Mandatory sub-agent delegation patterns to prevent 10K+ token context pollution.
177 GitHub stars

Casos de Uso

Speeding up iterative debugging and refactoring tasks using performance-optimized models like Grok.
Executing complex code implementations using high-reasoning models like GPT-5 or Gemini.
Conducting multi-model code reviews to cross-reference architectural suggestions and security patterns.

Acerca de

Características Principales

Access to 100+ OpenRouter models including Grok-3, GPT-5, and Gemini 2.0.
Support for both direct agent invocation and file-based instruction workflows.
Intelligent agent selection matrix for specialized tasks like architecture and code review.
Integrated CLI tools for model fuzzy searching, top-model recommendations, and cost tracking.
Mandatory sub-agent delegation patterns to prevent 10K+ token context pollution.
177 GitHub stars

Casos de Uso

Speeding up iterative debugging and refactoring tasks using performance-optimized models like Grok.
Executing complex code implementations using high-reasoning models like GPT-5 or Gemini.
Conducting multi-model code reviews to cross-reference architectural suggestions and security patterns.