A robust framework for managing multiple users, permissions, and isolated sandboxes within an AI agent environment.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install multi-user-workspace
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-user-workspace 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-User Workspace Management?
This skill provides a structured approach to multi-user management, enabling AI agents to handle diverse user profiles while maintaining strict data isolation. By leveraging Openclaw Skills, developers can create environments where every user session is uniquely identified, allowing for personalized interactions and role-based access control.
The core of this skill lies in its ability to manage sandboxed permissions and relationship awareness. It ensures that administrators maintain full visibility and control while guests or standard users operate within secure, isolated Docker containers, preventing unauthorized access to the broader workspace and maintaining privacy between different team members.
Multi-User Workspace Management Use Cases
- Providing isolated AI development environments for different team members within a shared server.
- Managing guest access to specific project directories with restricted tool permissions and Docker sandboxing.
- Building relationship-aware AI assistants that understand interpersonal dynamics and project collaborations.
- Enforcing security boundaries between administrator and guest roles in high-stakes shared workspaces.
How Multi-User Workspace Management Works
- The agent identifies the current user by extracting the userId from the active session key (e.g., agent:main:alice) during the initialization process.
- It references the USER.md registry to verify the user role, unique identifier, and specific permission levels.
- The system loads the corresponding user profile from the FRIENDS directory and relevant interpersonal data from the RELATIONS directory to establish context.
- Based on the role defined in the openclaw.json configuration, the agent initializes the session with either full access or an isolated Docker sandbox.
- Cross-user data boundaries are strictly enforced, allowing information sharing only when explicitly permitted in defined relationship files.
Multi-User Workspace Management Setup
Create the required directory structure for managing user profiles and relationships:
mkdir -p workspace/FRIENDS workspace/RELATIONS
Initialize the USER.md registry with user IDs and roles, and configure the openclaw.json file to map session regex patterns to specific agent configurations. For guest users, ensure their isolated directories are pre-created:
mkdir -p ~/.openclaw/workspace/guests/user_id
Finally, update your AGENTS.md to include instructions for session status checks and userId extraction to enable identity-aware greetings and data boundaries.
Multi-User Workspace Management Data Schema & Taxonomy
The skill organizes data through a specific file hierarchy to ensure clarity and security:
| Path | Purpose | Key Metadata |
|---|---|---|
| USER.md | Central registry of all users | userId, Name, Role |
| FRIENDS/ | Individual user profiles ({userId}.md) | Contact info, preferences, notes |
| RELATIONS/ | Interpersonal definitions ({u1}-{u2}.md) | Collaboration status, sharing rules |
| AGENTS.md | System instructions for identity logic | Identification steps, boundary rules |
| private/ | Admin-only restricted files | Sensitive configuration and logs |
Multi-User Workspace Management Advanced Features
- Docker-based session isolation with granular workspace mounting options per user.
- Automated session routing using regex pattern matching on session keys to assign correct agent profiles.
- Strict tool permission management through allow and deny lists based on user roles (e.g., disabling browser access for guests).
- Relationship-based information flow control that prevents data leaks while allowing collaborative context sharing when authorized.
SKILL.md
Loading
Related Openclaw Skills
Mission Control
A Next.js-powered command center for managing OpenClaw tasks, memories, and team activities through a visual web interface.
Futu Stock
A comprehensive stock market data skill for accessing real-time HK, US, and CN quotes, options, and account info via the Futu OpenAPI.
Skill Builder
A conversational guide that helps non-technical users design, refine, and export custom AI agent behaviors as ready-to-use SKILL.md files.
ClickUp Skill
A robust integration for the ClickUp project management platform, enabling automated task workflows and deep data analysis via REST API.
Openclaw Recovery Codex
A comprehensive diagnostic and recovery agent for Openclaw infrastructure, ensuring Gateway, channels, and pipelines remain operational across all platforms.
Prompt Request Pipeline
A fully automated GitHub pipeline that transforms issues tagged with [auto] into tested, reviewed, and merged pull requests.
Featured*
