An interactive framework for designing, configuring, and deploying multi-agent collaboration teams using standardized OpenClaw architecture.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-team-builder
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 agent-team-builder 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 Agent Team Builder?
Agent Team Builder is a specialized utility designed to guide developers through the complexity of multi-agent orchestration within the Openclaw Skills ecosystem. It facilitates the transition from single-agent setups to complex, multi-role teams by providing a step-by-step workflow for role definition, workspace isolation, and message routing. By following verified architecture patterns, it ensures that agents can collaborate effectively without session collisions or configuration errors.
This skill emphasizes the use of a single gateway process to host multiple specialized agents, each with its own identity and operational boundaries. It helps users implement the recommended orchestrator-specialist model, ensuring high-quality outputs and efficient task delegation. Whether you are building a software development squad or a content creation team, this skill provides the templates and logic needed for robust multi-agent Openclaw Skills implementations.
Agent Team Builder Use Cases
- Creating a full-stack software development team with dedicated roles for planning, implementing, and reviewing code.
- Setting up a multi-agent business operations unit to handle research, analysis, and executive reporting.
- Deploying a coordinated content creation pipeline where different agents handle strategy, drafting, and quality control.
- Configuring high-performance group collaboration environments on Discord or Telegram using specialized Openclaw Skills agents.
How Agent Team Builder Works
- Conduct a Team Design phase to define agent IDs, display names, and specific roles like orchestrators or specialists.
- Perform Architecture Planning to map agents to a single gateway and select appropriate communication channels.
- Execute Agent and Workspace Setup by generating standardized Markdown files like AGENTS.md and SOUL.md for each team member.
- Configure Routing and Bindings to ensure incoming messages from Discord, Telegram, or WhatsApp reach the correct agent.
- Establish Collaboration Rules including mention patterns and ping-pong safety limits to prevent infinite loops.
- Enable Agent-to-Agent Communication using the sessions_send and sessions_spawn primitives for inter-agent delegation.
- Implement Team Shared Memory using symlinked directories to allow cross-agent knowledge sharing without token bloat.
- Finalize Memory and Operations settings to optimize model tiering, cost control, and heartbeat checks.
Agent Team Builder Setup
To begin building your team with Openclaw Skills, use the CLI to initialize your agents:
openclaw agents add <agent-id>
Once initialized, define your agent list in your configuration file located at ~/.openclaw/openclaw.json to specify workspace paths and model preferences for each team member.
Agent Team Builder Data Schema & Taxonomy
The skill utilizes a structured file-based system to manage agent state and team logic. Each agent operates within a dedicated workspace following this schema:
| File Type | Filename | Purpose |
|---|---|---|
| Configuration | openclaw.json |
Global bindings, account tokens, and agent registry. |
| Persona | SOUL.md |
Defines the agent's tone, boundaries, and communication style. |
| Operations | AGENTS.md |
Core operating instructions and memory protocols. |
| Shared | team-shared/ |
Symlinked directory containing cross-agent status and decision logs. |
| Memory | MEMORY.md |
Long-term durable facts, loaded exclusively in private sessions. |
| Logs | memory/YYYY-MM-DD.md |
Daily append-only logs used for semantic search indexing. |
Agent Team Builder Advanced Features
- Synchronous inter-agent communication via sessions_send with configurable turn limits.
- Asynchronous task delegation through sessions_spawn for complex sub-agent workflows.
- Orchestrator-level session visibility allowing a lead agent to monitor the entire communication tree.
- Advanced mention patterns and group policy allowlists for granular control in collaborative channels.
- Model tiering capabilities to leverage expensive models for reasoning and cheaper models for execution within Openclaw Skills.
- Symbolic link-based shared memory architecture for near-instant cross-agent context propagation.
SKILL.md
Loading
Related Openclaw Skills
Fullstack Project Creator
Automates the creation of standardized Java Spring Boot and Vue3 full-stack project structures with integrated database scripts and Docker configuration.
WeChat Article Reader (weixin-reader-oc)
A high-performance extraction tool for retrieving full text and metadata from WeChat Public Account articles without requiring a login.
Oura Ring CLI
A command-line tool for retrieving comprehensive health and biometric data from the Oura Ring API (V2).
Vue Component Generator
A high-speed CLI tool for generating standardized Vue 3 components with support for TypeScript, SCSS, and multiple API styles.
Tarot Skill
A professional-grade tarot reading module that enables AI agents to conduct deep-dive spreads, provide energy guidance, and generate visual card interpretations.
Dida365 (TickTick) Task Manager
A comprehensive CLI-based tool for managing Dida365 tasks and projects through Openclaw Skills.
Featured*
