Skip to content

UAIX vs MCP vs Webhooks

UAIX, MCP, and webhooks answer different questions. UAIX records the validated exchange and evidence boundary. MCP connects a model client to tools, resources, prompts, or application context. Webhooks deliver event notifications between systems.

Quick chooser

Need Use Reason
Portable public record, release evidence, validation, provenance, trust, or conformance. UAIX / UAI-1 The output needs to be reviewed, cited, replayed, or attached to a support claim.
Tool, resource, or prompt access inside a model-client session. MCP The session needs runtime access, not necessarily a durable public evidence packet.
One system notifying another that an event happened. Webhook The main requirement is event delivery, not a full standards envelope.
Runtime action plus public evidence. Use together Let the runtime/tool/event layer do the work, then export the reviewable boundary as UAI-1 evidence.

Capability-adaptive rule

A client may use MCP tools or receive webhook events and still publish a UAIX capability profile. The profile should say which level the client actually supports, what fallback applies, what consent is required, and what evidence must travel.

References inside UAIX