How do I install the necessary tools for this skill?

You must install the Claudish CLI globally using 'npm install -g claudish' and have a valid OpenRouter API key set in your environment variables.

Why should I run Claudish through a sub-agent instead of the main context?

Running Claudish directly pollutes the main context window with over 10,000 tokens of reasoning and tool outputs. Sub-agents isolate this data, keeping your main conversation efficient.

Does Claudish support all Claude Code features?

Yes, Claudish acts as a proxy that supports 100% of Claude Code’s capabilities while allowing you to switch the underlying model provider through OpenRouter.

What is the file-based instruction pattern?

It is a best practice where instructions are written to a temporary markdown file and passed to Claudish via stdin, ensuring complex multi-step workflows are handled cleanly by sub-agents.

Which models are recommended for coding tasks within Claudish?

Grok-code-fast-1 is excellent for speed and iterations, while Gemini 1.5 Pro and GPT-4o are recommended for complex reasoning and architecture planning.

Claudish Model Integration

Name: Claudish Model Integration
Author: tianzecn

bytianzecn

•

Herramientas para Desarrolladores

Enables Claude Code to interact with external LLMs via OpenRouter while optimizing context window efficiency through specialized sub-agent delegation.

Acerca de

This skill provides a comprehensive framework for using the Claudish CLI to run Claude Code with alternative models like Grok, GPT-5, and Gemini via OpenRouter. It prioritizes performance and context management by mandating a sub-agent execution pattern, preventing the main conversation from becoming bloated with reasoning tokens. The skill includes an agent selection matrix, automated environment setup guides, and best practices for file-based instruction passing to ensure complex coding tasks are handled by the most capable models without degrading the user's primary workspace.

Características Principales

Access to 100+ OpenRouter models including Grok, Gemini, and GPT-5
Built-in agent selection matrix for matching models to specific tasks
Context-safe sub-agent delegation pattern to prevent token pollution
Direct agent invocation via the --agent flag for streamlined workflows
1 GitHub stars
Interactive and single-shot CLI modes with cost tracking

Casos de Uso

Using Gemini’s massive context window for deep codebase analysis and investigation
Comparing model outputs for complex architectural decisions or refactoring tasks
Leveraging Grok’s high-speed reasoning for rapid code iterations

Acerca de

Características Principales

Access to 100+ OpenRouter models including Grok, Gemini, and GPT-5
Built-in agent selection matrix for matching models to specific tasks
Context-safe sub-agent delegation pattern to prevent token pollution
Direct agent invocation via the --agent flag for streamlined workflows
1 GitHub stars
Interactive and single-shot CLI modes with cost tracking

Casos de Uso

Using Gemini’s massive context window for deep codebase analysis and investigation
Comparing model outputs for complex architectural decisions or refactoring tasks
Leveraging Grok’s high-speed reasoning for rapid code iterations