A collaborative dual-agent workflow that pairs a Project Manager agent with a Developer agent to architect and build software projects.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install multi-agent-dev-team
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install multi-agent-dev-team using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Multi-Agent Dev Team?
The Multi-Agent Dev Team skill transforms the way developers interact with AI by implementing a structured, hierarchical team environment within Openclaw Skills. Instead of interacting with a single general-purpose assistant, this skill provides a specialized Project Manager (PM) agent and a Developer (Dev) agent that work in tandem. The PM agent acts as the orchestrator, translating your high-level goals into technical specifications, while the Dev agent focuses on the implementation, testing, and version control of the code.
By leveraging this multi-agent architecture, the skill ensures higher quality outputs and better project organization. It is designed to handle the complexities of modern web development, allowing the AI to maintain context over planning and execution phases separately. This approach significantly reduces errors and produces cleaner, more maintainable codebases for prototypes and web applications.
Multi-Agent Dev Team Use Cases
- Rapidly scaffolding Next.js and React applications with Tailwind CSS.
- Building functional MVPs and prototypes from simple text descriptions.
- Generating reusable TypeScript component libraries and API integrations.
- Managing complex coding tasks that require separate planning and execution phases.
- Implementing best-practice multi-agent workflows within the Openclaw Skills ecosystem.
How Multi-Agent Dev Team Works
- The user describes the project requirements to the PM agent acting as the Director.
- The PM agent analyzes the requirements and creates a structured Task Specification document.
- The PM agent spawns the Dev agent and provides it with the technical constraints and acceptance criteria.
- The Dev agent executes the coding tasks, initializes a Git repository, and writes the necessary files.
- The PM agent reviews the generated code and deliverables to ensure they meet the project goals.
- The PM agent reports completion to the user or initiates up to three iterations of improvements if needed.
Multi-Agent Dev Team Setup
First, install the skill using the CLI:
npx clawhub install multi-agent-dev-team
Next, configure the agents in your ~/.openclaw/config.yaml file to link the SOUL definitions to your preferred LLM providers:
agents:
multi-agent-pm:
soul: ~/.openclaw/skills/multi-agent-dev-team/agents/pm-agent/SOUL.md
model: anthropic/claude-sonnet-4-5-20250929
multi-agent-dev:
soul: ~/.openclaw/skills/multi-agent-dev-team/agents/dev-agent/SOUL.md
model: google/gemini-2.5-flash
Start the workflow by chatting with the PM agent:
openclaw chat --agent multi-agent-pm
Multi-Agent Dev Team Data Schema & Taxonomy
The Multi-Agent Dev Team skill organizes project data into a clean, professional directory structure. It utilizes specific metadata and templates to maintain state across Openclaw Skills sessions:
| Component | Format | Purpose |
|---|---|---|
| Project Workspace | Directory | Contains all source code, typically under ~/projects/ or a custom cwd. |
| Task Specification | Markdown | A structured hand-off document containing requirements and acceptance criteria. |
| Version Control | Git | Every project is initialized with a Git repository for change tracking. |
| Session History | JSON/YAML | Tracks the communication logs and iteration count between the PM and Dev agents. |
Multi-Agent Dev Team Advanced Features
- Cross-Model Optimization: Use a high-reasoning model for the PM and a high-speed model for the Dev agent.
- Automated Iteration: The PM agent can automatically trigger up to 3 revision cycles based on quality checks.
- Custom Workspace Routing: Define specific directories for project creation using the
cwdconfiguration option. - Structured Hand-offs: Uses standardized Markdown templates for inter-agent communication to ensure consistency.
- Pro Upgrade Path: Access specialized agents for QA, DevOps, and Architecture for enterprise-grade Openclaw Skills workflows.
SKILL.md
Loading
Related Openclaw Skills
VeriGlow Agent Map
A documentation registry that enables AI agents to discover internal APIs and browser automation recipes for any website.
WordPress API with Gutenberg Integration
A comprehensive automation toolkit for managing WordPress content via REST API with native Gutenberg block support.
Bounty Hunter
An automated smart contract vulnerability scanner that leverages Slither and local LLMs for efficient bug bounty hunting.
Language Learning Tutor
Transform your AI agent into a polyglot language tutor with customizable lessons, flashcards, and real-world conversation practice.
Plausible Analytics
An AI-ready interface for querying privacy-focused website metrics and performance data via the Plausible Analytics API.
Japanese Tutor
An interactive AI language assistant that automates Japanese study through document parsing, OCR translation, and pedagogical guidance.
Featured*
