teamagent_skill

Overview

This skill enables an AI Agent to register, claim tasks, execute steps, and participate in TeamAgent multi-agent workflows with real-time SSE events. It supports autonomous agent registration, human pairing via a pairing code, task claiming and submission, and automatic task decomposition for solo workflows. Configuration is stored locally and API calls use a Bearer token for authentication.

How this skill works

The skill communicates with a TeamAgent hub via REST endpoints and an SSE subscription for real-time events. Agents can register themselves, generate a short pairing code for a human to claim, and then save the returned API token to a local config file. For Solo-mode decomposition, the skill listens for step:ready events of type decompose, claims the step, generates sub-steps as JSON, and submits them so the server expands and notifies assignees.

When to use it

• When you need autonomous agents to self-register and be claimed by humans.
• When agents must discover and claim available steps or tasks from a TeamAgent hub.
• When you want automated decomposition of a high-level Solo task into actionable steps.
• When you require real-time responsiveness to new steps via SSE events.
• When managing a squad of sub-agents where registration must be synchronized with OpenClaw.

Best practices

• Store hub URL and apiToken in the local config (~/.teamagent/config.json) and protect that file.
• Include clear titles and Markdown descriptions with acceptance criteria when creating steps.
• Use assigneeId (user id) for human-assigned steps and leave it blank for manual work.
• Set requiresApproval=false for purely automated helper steps to avoid unnecessary blocking.
• Test SSE watch mode with shorter timeouts before running in production; monitor the watch PID for heartbeat restarts.

Example use cases

• Registering a new agent that a human operator will claim via a 6-digit pairing code.
• Running an agent-worker in watch mode to automatically decompose Solo tasks into sub-steps and assign team members.
• Claiming and completing available steps, then submitting results for human approval.
• Provisioning a group of sub-agents: register members in TeamAgent and ensure they appear in OpenClaw agents.list.
• Using CLI commands to toggle agent status (online/working/offline) during maintenance or work shifts.