Multi-Agent Dev Team

What This Skill Does

The Multi-Agent Dev Team skill is a sophisticated workflow automation for OpenClaw that replicates a professional engineering team environment. By leveraging two specialized agents, it automates the end-to-end software development lifecycle. The Project Manager (PM) agent acts as an orchestrator, translating natural language requirements into technical task specifications, while the Dev agent functions as the implementation engine. This architecture ensures that code generation is structured, requirements-driven, and aligned with project constraints, effectively reducing the cognitive load on the user.

Installation

To install this skill, use the ClawHub command-line tool. Execute the following command in your terminal: npx clawhub install multi-agent-dev-team. After installation, navigate to your OpenClaw configuration file located at ~/.openclaw/config.yaml to assign specific models to the multi-agent-pm and multi-agent-dev agents. It is recommended to use a powerful reasoning model like Claude 3.5 Sonnet for the PM agent to handle complex task decomposition, and a fast, high-token-throughput model like Gemini 2.5 Flash for the Dev agent to accelerate code synthesis.

Use Cases

• Rapid Prototyping: Quickly scaffold MVPs for web applications to validate business ideas without manual boilerplate setup.
• Content-Driven Websites: Build professional landing pages, portfolios, or documentation sites using modern frameworks like Next.js.
• Automated Code Tasks: Delegate repetitive coding assignments, such as creating API routes, unit tests, or UI components, to the Dev agent.
• Technical Learning: Observe how structured agentic workflows handle project planning, error checking, and Git integration.

Example Prompts

1. "Build me a personal task management dashboard using React and Tailwind CSS. Include a sidebar for categories and a persistent local storage layer."
2. "Create a responsive product landing page for a coffee subscription service. Focus on a strong hero section, a pricing comparison grid, and a sign-up form that validates email input."
3. "Develop a Python script that scrapes headlines from three major tech news sites and saves them into a formatted markdown file on my desktop."

Tips & Limitations

• Task Granularity: The PM agent works best when you provide clear, objective-based requirements. If a task is too vague, the PM may struggle to define constraints.
• Iteration Limits: By default, the workflow supports a maximum of 3 refinement iterations. Use this cycle wisely to steer the code generation toward your vision.
• Security Awareness: Because this skill performs file-write and code-execution, always review the generated code before running it in a production or sensitive environment.
• Model Selection: Pair the PM agent with high-reasoning models to improve the quality of the task breakdown, which significantly impacts the output of the Dev agent.