OAuth2.0 Extension for Multi-AI Agent Collaboration

This draft proposes an authorization method for task-oriented multi- AI agent collaboration scenarios, where a leading agent coordinates sub-agents to complete complex tasks. The methods extends the OAuth 2.0 protocol by adding fields in token and message flows, enabling sub-agents execute the task as a group. This approach greatly simplifies the authorization process for a task group, avoids efficiency loss from repeated interactions between multiple sub-AI agents and the authorization server, meanwhile maintaining full compatibility with existing OAuth 2.0 workflows.