What is the Proxy Mode Pattern?

It is a specialized workflow that allows a Claude agent to delegate specific tasks to external AI models (like Grok) using the Claudish CLI, returning an attributed response to the main session.

Which tools are used for the quality check patterns?

The patterns are pre-configured to work with common modern tools like Biome.js for linting and formatting, TypeScript for type-checking, and Vitest for testing.

Can I use these patterns for automated code reviews?

Yes, the skill includes a Review Feedback Pattern that provides a structured format for PASS/FAIL decisions, issue categorization, and impact analysis.

How does this skill improve task tracking?

It implements a TodoWrite requirement that forces agents to initialize a task list and update the status of each phase (in_progress/completed) as they work.

Agent Development Patterns

Name: Agent Development Patterns
Author: involvex

byinvolvex

生产力与工作流

Provides standardized templates and architectural patterns for building robust, multi-agent systems within Claude Code.

关于

Agent Development Patterns is a comprehensive blueprint library designed for developers building autonomous agents. It offers proven implementation patterns for delegating tasks to external AI models via Proxy Mode, managing workflow state through TodoWrite, and enforcing automated quality gates like linting and testing. By providing standardized templates for planners, implementers, and reviewers, this skill helps ensure that custom Claude agents are consistent, reliable, and capable of complex multi-step orchestration.

主要功能

Structured review feedback templates for consistent agent-to-agent validation
Automated quality check gates for formatting, linting, and type-checking
0 GitHub stars
Standardized TodoWrite workflow tracking for real-time progress monitoring
Pre-defined YAML templates for specialized Planner, Implementer, and Reviewer roles
Proxy Mode delegation for external AI model integration via Claudish

使用场景

Integrating non-Claude models into a local workflow for specific coding tasks
Enforcing strict CI/CD quality standards for code generated by autonomous agents
Building a multi-agent system where an Orchestrator delegates to specialized sub-agents

关于

主要功能

Structured review feedback templates for consistent agent-to-agent validation
Automated quality check gates for formatting, linting, and type-checking
0 GitHub stars
Standardized TodoWrite workflow tracking for real-time progress monitoring
Pre-defined YAML templates for specialized Planner, Implementer, and Reviewer roles
Proxy Mode delegation for external AI model integration via Claudish

使用场景

Integrating non-Claude models into a local workflow for specific coding tasks
Enforcing strict CI/CD quality standards for code generated by autonomous agents
Building a multi-agent system where an Orchestrator delegates to specialized sub-agents