What it does

The Agent Coordination Protocol is a tool-agnostic operating procedure designed to transform a single AI agent into a disciplined orchestrator of specialized sub-agents. It eliminates the chaos common in multi-agent workflows—such as drifting objectives, duplicated work, and contradictory outputs—by enforcing a rigorous system of role contracts and structured handoffs.

Why use this skill

Managing multiple agents often leads to "context rot" where critical information is lost during transitions. This skill provides the mental framework for an AI to decompose complex objectives into clear, non-overlapping roles. Instead of ad-hoc prompting, it uses a deterministic model of Objective Contracts and Handoff Packets to ensure every sub-agent knows exactly what they own and what they must not touch.

Supported tools

Because this is a reasoning protocol rather than code, it works with any agent that reads the SKILL.md standard, including Claude Code, Codex CLI, Cursor, VS Code Copilot, and Gemini CLI. It operates entirely through the agent's reasoning and structured output, with no servers, scripts, or external dependencies.

The output

The skill forces the agent to produce transparent coordination artifacts: a Shared Context Ledger of decisions, explicit success criteria, and structured packets when passing tasks between roles. This ensures the human user can audit the "why" and "how" of a collaborative AI workflow.