A technical framework for deploying multiple specialized AI agents in a single Telegram group with a unified workspace and shared memory.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install huamu668-openclaw-multiagent
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 huamu668-openclaw-multiagent 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 Openclaw Multi-Agent Telegram System?
The multi-agent system for Telegram allows developers to orchestrate a team of specialized AI agents within a single group chat. By utilizing the Openclaw Skills architecture, this configuration enables a main agent to coordinate with various sub-agents through a shared workspace directory. This setup is ideal for complex workflows where different bots handle specific roles while maintaining a consistent project state.
At the core of this skill is a shared-context model where all agents can read from and write to a centralized file system. This ensures that the collective intelligence of the group is synchronized. Furthermore, the integration of MemOS Cloud removes the need for manual logging, providing a seamless global memory pool that all agents in the Openclaw Skills ecosystem can access and contribute to automatically.
Openclaw Multi-Agent Telegram System Use Cases
- Deploying a specialized AI workforce in a project-specific Telegram group.
- Creating a hierarchy of agents where a manager bot delegates tasks to specialized sub-bots.
- Enabling collaborative cross-agent workflows that require shared context and history.
- Managing complex technical projects on the go using multiple bot interfaces within one chat interface.
How Openclaw Multi-Agent Telegram System Works
- The user establishes a shared workspace directory structure that serves as the single source of truth for all agents.
- Sub-agents are assigned unique identities and specialized behavioral rules through localized SOUL and AGENTS configuration files.
- Telegram bot privacy settings are disabled to allow the bots to read and respond to the group's message flow.
- The system configuration is migrated to a multi-account mode, binding specific agent IDs to unique Telegram bot tokens.
- Agents communicate and collaborate by reading from and writing to the shared-context directory for real-time signaling.
- Global memory is managed via MemOS Cloud, ensuring that context is preserved across all agents without duplicating data.
Openclaw Multi-Agent Telegram System Setup
1. BotFather Configuration
For each sub-agent bot, you must message @BotFather, use /setprivacy, and select Disable. This allows the bots to read group messages.
2. Directory Initialization
Prepare the shared workspace and agent directories:
# Create shared context
mkdir -p ~/.openclaw/workspace/shared-context
# Create sub-agent workspace and internal data folders
mkdir -p ~/.openclaw/workspace/agents/{agent_id}
mkdir -p ~/.openclaw/agents/{agent_id}/agent
3. Configuration Update
Backup your existing config and update openclaw.json to enable accounts mode in the Telegram channel section, ensuring agentToAgent tools are active.
Openclaw Multi-Agent Telegram System Data Schema & Taxonomy
| Path | Description |
|---|---|
workspace/SOUL.md |
Core identity and goals of the Primary Agent. |
workspace/shared-context/ |
Global layer for FEEDBACK-LOG and SIGNALS accessible by all agents. |
workspace/agents/{id}/ |
Dedicated directory for sub-agent specific logic and memory files. |
openclaw.json |
Master configuration mapping bot accounts to specific agent identities. |
workspace/USER.md |
Shared user profile and preference data for consistent agent interaction. |
Openclaw Multi-Agent Telegram System Advanced Features
- Shared Memory Architecture: Uses MemOS Cloud to provide a unified memory pool across the entire agent team.
- File-Based Signaling: Agents can trigger collaborative actions by writing to shared SIGNALS.md files.
- Multi-Account Binding: Support for distinct bot identities within the same Telegram group for role-based interactions.
- Granular Privacy Control: Configuration options to require mentions for sub-agents while allowing the main agent to listen to all traffic.
- Robust Backup Logic: Integrated steps for state recovery and configuration versioning.
SKILL.md
Loading
Related Openclaw Skills
MemOS Memory Recall
A persistent memory skill that enables AI agents to query and recall past interactions and user preferences from MemOS Cloud.
Best Practices
Enforce modern web development standards, security protocols, and code quality patterns within your AI-driven development environment.
Agent OS
A structural framework for AI agents that separates identity, operational logic, and knowledge management for better scalability and consistency.
News Aggregator
An automated news intelligence agent that gathers, filters, and summarizes global technology and military news from authoritative sources.
OpenClaw Security Practice Guide
A comprehensive security framework for AI agents with root access, covering lifecycle auditing, permission hardening, and automated disaster recovery.
Performance Optimization (Web & LLM Cost)
A comprehensive optimization toolkit for boosting web application speed and minimizing LLM API expenditures through intelligent resource management.
Featured*
