Define Agent Exchange as the directory/control-plane layer built on Relay.
Working definition:
Agent Exchange is the Helpthread layer that lets humans, assistants, systems, and external agents discover eligible participants, understand capabilities, request work, and collaborate through Relay with explicit authority and provenance.
It is not the Relay lifecycle itself. Relay is the work/request mechanism. Agent Exchange is the actor, capability, trust, and routing surface that uses Relay.
Scope:
- actor registry
- capability catalog
- trust and grant model
- Relay routing rules
- audit/provenance expectations
- protocol-neutral adapters such as A2A, MCP, webhooks, REST, or message-native commands
Non-goals:
- global public marketplace
- hosted broker
- cross-tenant network operated by Helpthread or Resonant IQ
- automatic internet-wide agent discovery
- payment/settlement between agents
- reputation scoring
- broad autonomous access
Acceptance:
- Public RFC defines what Agent Exchange is and is not.
- RFC explains how Agent Exchange differs from Relay.
- RFC maps Agent Exchange onto existing Helpthread primitives: Agents, Assistants, conversations, events, webhooks, drafts, and Relay.
- RFC includes examples for vendor escalation, internal approval, assistant-mediated action, and message-originated approval.
- RFC preserves the charter boundary: no hosted core, no Resonant IQ data path, no private first-party privileges.
Define Agent Exchange as the directory/control-plane layer built on Relay.
Working definition:
Agent Exchange is the Helpthread layer that lets humans, assistants, systems, and external agents discover eligible participants, understand capabilities, request work, and collaborate through Relay with explicit authority and provenance.
It is not the Relay lifecycle itself. Relay is the work/request mechanism. Agent Exchange is the actor, capability, trust, and routing surface that uses Relay.
Scope:
Non-goals:
Acceptance: