Why shouldn't I run Claudish in the main context?

Running Claudish directly in the main context can add over 10,000 tokens of conversation history and reasoning to your window, leading to context pollution and inefficiency.

Which models are supported by Claudish?

Claudish supports any model available through OpenRouter, including Grok, Gemini, GPT-5, MiniMax, and Qwen.

What is the sub-agent delegation pattern?

It is a workflow where Claudish tasks are assigned to a secondary agent or instruction file, isolating the heavy token usage from your primary chat session.

Do I need a separate API key for this skill?

Yes, you must have an OpenRouter API key configured as an environment variable (OPENROUTER_API_KEY) to use external models.

Can I use Claudish for code reviews?

Absolutely. Claudish can delegate code review tasks to specialized agents using high-reasoning models like GPT-5 or Gemini for thorough analysis.

Claudish Multi-Model Integration

Name: Claudish Multi-Model Integration
Author: MadAppGang

byMadAppGang

•

177

•

Data Science & ML

Integrates Claude Code with OpenRouter models like Grok, GPT-5, and Gemini while maintaining context efficiency through sub-agent delegation.

The Claudish Usage skill enables developers to leverage the full power of OpenRouter's extensive model catalog directly within their Claude Code environment. It provides a standardized framework for proxying requests to alternative LLMs—such as Grok, Gemini, and GPT-5—while strictly enforcing a sub-agent delegation pattern. This architecture prevents the main conversation context from being overwhelmed by large token counts, ensuring that multi-model workflows remain high-performance, cost-effective, and manageable even during complex, multi-phase development tasks.

Key Features

01Multi-model support for Grok, GPT-5, Gemini, and MiniMax via OpenRouter proxying

02File-based instruction patterns for isolated, production-grade agent workflows

03177 GitHub stars

04Comprehensive agent selection matrix for specialized coding, testing, and architecture

05Integrated model discovery and curated 'Top Models' list for development

06Strict sub-agent delegation rules to prevent main context window pollution

Use Cases

01Delegating specialized frontend or backend implementations to domain-specific sub-agents

02Performing deep architectural reviews or complex reasoning tasks using GPT-5

03Running rapid code iterations using speed-optimized models like Grok-Code-Fast

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add madappgang/claude-code claudish-usage

For use in Claude.ai and ChatGPT

Download Skill