LaunchKit · 2026
Back to Skills

agents-manager

Manage Clawdbot agents: discover, profile, track.

5
3.4k downloads
by @agentandbot-design

Setup & Installation

openclaw skills install @agentandbot-design/agents-manager

Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:

npx clawhub install agents-manager

Version History

v1.0.2Jan 31, 2026 - Version note by skill developer:

**Summary: Modernized and expanded agent management with new scripts, protocols, and clearer human/agent workflows.** - Added comprehensive agent management scripts (discovery, validation, health checking, stats, visualization, setup). - Introduced "Handshake Protocol" and approval workflows for task assignment. - Replaced older references/structure files to fix naming consistency. - Split guidance for humans and agents, including step-by-step commands and protocols. - Included detailed references and agent registry schema for easier onboarding and management. - Updated metadata and homepage for better integration and discoverability.

What This Skill Does

Centrally manages Clawdbot agents across a multi-agent system. Handles discovery, capability profiling, routing hierarchy, and task assignment with approval-aware handshake protocols.

Centralizes agent discovery, permission checks, and routing rules in one place instead of hardcoding agent relationships across individual scripts.

When to use it

  • Discovering all agents available in a Clawdbot deployment
  • Checking whether one agent is permitted to assign tasks to another
  • Validating agent registry integrity before a deployment
  • Visualizing the full agent hierarchy and reporting structure
  • Routing a new task through the escalation chain to the correct agent

Example Workflow

Here's how your AI assistant might use this skill in practice.

INPUT

User asks: Route an incoming task to the right agent in my Clawdbot system

AGENT
  1. 1Run scan_agents.js to list all available agents
  2. 2Check agent profiles to identify a capability match for the task
  3. 3Verify permissions between source and target agents with can_assign.js
  4. 4Check the requires_approval flag in agent-registry.md
  5. 5Assign task directly if approval is not required, or escalate to supervisor
OUTPUT

Task routed to the correct agent, or an approval request sent up the escalation chain