📄️ Agent Identity
This domain curates everything that makes an agent recognizable: canonical metadata, ownership signals, and brand assets. It does not run inference; it simply keeps the public-facing profile accurate and reproducible.
📄️ Agent Knowledge
This domain keeps each agent’s long-term memory. It provisions private databases, stores prompt templates and conversation contexts, and serves them back to the reasoning stack before every inference. Nothing is on-chain; all state lives in the agent’s own database.
📄️ Agent Reasoning
This domain is the agent’s thinking stack. It composes prompts from the latest context, talks to LLM providers, triggers MCP tools, and returns responses that channels can send back to humans.
📄️ Channels & Integrations
This domain gives agents a way to talk to humans where they already collaborate (Slack today, more channels later). Think of it as the translation layer between external workspaces and the orchestrator: it receives events from a channel, keeps enough local state to understand the conversation, and sends the agent’s reply back out.