What is the 2-agent workflow?

It is a collaborative system where Cursor acts as the Project Manager (PM) for high-level planning and review, while Claude Code acts as the Worker for implementation, testing, and staging.

Does this skill execute code changes?

No, this is a guidance skill designed to explain the workflow and process. It does not perform implementation work or workflow setup itself.

How are tasks managed in this workflow?

Tasks are managed through a Plans.md file using specific markers like pm:依頼中 (Requested), cc:WIP (In Progress), and cc:完了 (Completed) to track state.

What is the CI/CD 3-strike rule?

If Claude Code fails to fix a CI error after three consecutive attempts, it automatically stops, generates an escalation report, and asks the PM (Cursor) for guidance.

Agentic Workflow Guide

Name: Agentic Workflow Guide
Author: Chachamaru127

byChachamaru127

•

676

•

Productivity & Workflow

Guides developers through the collaborative 2-agent workflow between Cursor and Claude Code for autonomous development cycles.

This skill provides specialized guidance for the 'Plan-Work-Review' cycle, orchestrating a high-efficiency environment where Cursor acts as the Project Manager (PM) and Claude Code serves as the Worker. It defines clear roles, provides standardized task management protocols using Plans.md status markers, and outlines command structures for seamless agent-to-agent handoffs. Ideal for teams looking to implement autonomous coding workflows, it ensures that implementation, testing, and staging deployments follow a structured, predictable path while preventing common pitfalls in AI-driven development.

Key Features

01Standardized task tracking system using specific Plans.md status markers.

02Detailed guidance on /handoff and /work command sequences.

03Visual state transition logic from task assignment to production review.

04676 GitHub stars

05Automated CI/CD '3-strike' rule for intelligent error escalation.

06Comprehensive role definitions for Cursor (PM) and Claude Code (Worker) agents.

Use Cases

01Establishing a multi-agent AI development environment with clear accountability.

02Troubleshooting agent collaboration and CI/CD escalation protocols.

03Learning how to manage complex task states across different AI coding tools.

Key Features

01Standardized task tracking system using specific Plans.md status markers.

02Detailed guidance on /handoff and /work command sequences.

03Visual state transition logic from task assignment to production review.

04676 GitHub stars

05Automated CI/CD '3-strike' rule for intelligent error escalation.

06Comprehensive role definitions for Cursor (PM) and Claude Code (Worker) agents.

Use Cases

01Establishing a multi-agent AI development environment with clear accountability.

02Troubleshooting agent collaboration and CI/CD escalation protocols.

03Learning how to manage complex task states across different AI coding tools.