◇ 未来
Universal Memory API
A portable memory protocol for AI agents.
memoryagentscontext
A portable memory protocol for AI agents.
A standard identity layer for proving which autonomous agent is acting.
A proposed protocol for securely transferring structured context between autonomous agents.
A bounded service for investigating, fixing, and verifying software defects.
A user-controlled container for portable agent memories.
A machine-readable representation of an agent's verified intention.
A neutral context layer for moving state between model families.
A standard way for agents to request and receive tool permissions.
A bounded agent that prepares, verifies, and releases software changes.
A permissioned wallet system allowing agents to transact within defined budgets.
A portable authorization and capability record for MCP clients and servers.
A browser substrate designed for agent-native interaction.
A semantic bridge between model-specific prompts, tools, and responses.
A machine-readable consent layer for personal and organizational data used by AI systems.
A semantic compression service for long-running agent context.
A resumable scheduling layer for long-running agent tasks.
A portable reputation system for autonomous agents and the operators behind them.
A graph-native representation of durable agent knowledge and relationships.
A structured communication protocol for agents working across organizations.
A durable identity for work that survives model, session, and infrastructure changes.
An authorization flow based on what an agent intends to accomplish, not only which API it may call.
A portable market for discovering, installing, and governing reusable agent skills.
A structured interface for agents to negotiate constraints, deadlines, and tradeoffs.
A version control and evaluation registry for production prompts and agent instructions.
A declared interface between websites and agent browsers.
A structured catalog of model behavior profiles for predictable product design.
A contextual trust credential derived from verified agent and system behavior.
A controlled environment for agents to use isolated, auditable identities.
A governed market for matching bounded digital work with autonomous workers.
A proposed risk-sharing layer for bounded autonomous operations.