agent-teamslisted

Orchestrate autonomous multi-agent teams using Captain-led delegation, durable subagents, dependency-aware task DAGs, peer-to-peer messaging, and structured quality gates. Activate when the user asks to use agent-teams, build features with an agent team, or for complex multi-stage engineering tasks and TDD workflows.
sanad-source/agent-teams · ★ 1 · AI & Automation · score 74

Install: claude install-skill sanad-source/agent-teams

# AgentTeams: Multi-Agent Team Orchestration Skill The **AgentTeams** skill transforms Antigravity from a single-threaded coding assistant into a **Captain** orchestrating a coordinated, specialized team of autonomous subagents. Inspired by the [dsh-agent-teams](https://github.com/NanmiCoder/dsh-agent-teams) protocol, this skill enforces structured delegation, clean separation of concerns, dependency-aware task graphs (DAGs), direct peer-to-peer agent messaging, and machine-verifiable **Quality Gates** (requirements → implementation → verification → review → repair → integration). --- ## How to Call / Activate This Skill You can activate this skill in any of the following ways: ### 1. Explicit Skill Invocation (Recommended) Mention the skill name directly in your request or slash command: - `"Use the agent-teams skill to implement <feature>"` - `"/goal use agent-teams to refactor <module>"` - `"Activate agent-teams: build <system> with a full review cycle"` ### 2. Natural Language Request Ask for a team-based or multi-agent execution: - `"Assemble an agent team to design, code, and test this"` - `"Coordinate a multi-agent workflow with QA and independent review"` - `"Use subagents as a team with a task DAG to build <feature>"` ### 3. Teamwork Slash Command Trigger teamwork planning via slash command: - `/teamwork-preview` --- ## 1. Core Principles 1. **Strict Model Inheritance**: - **All agents and subagents MUST use `Model: 'inherit'`.** - This ensures every