Skip to content
NimbleBrainNimbleBrainNimbleBrain

Protocol Support

This page is the contract a server developer builds against: what the host speaks, what it claims in the handshake, and what it leaves out on purpose. Where the host does not advertise a capability, that is a decision, and the reason is given next to it.

Specification revisions

Section titled “Specification revisions”

NimbleBrain uses the official TypeScript SDK (@modelcontextprotocol/sdk 1.x) for both its client and server roles. Version negotiation follows the specification’s lifecycle rules: the host offers the newest revision it supports, and it agrees to any older supported revision the other side asks for.

Revision Supported
2025-11-25 Yes. This is the revision the host offers first.
2025-06-18 Yes
2025-03-26 Yes
2024-11-05 Yes
2024-10-07 Yes

MCP Apps (the ui:// extension) is implemented from @modelcontextprotocol/ext-apps. See the MCP App Bridge for the exact messages the host serves.

As a client to your server

Section titled “As a client to your server”

Transports

Section titled “Transports”
Transport Supported Notes
Streamable HTTP Yes The default. Declare streamable-http in the server’s catalog entry.
HTTP+SSE Yes Only for servers that support nothing newer. Declare sse.
stdio No The runtime connects to servers over the network. It never downloads or runs a server’s code.

Authorization follows the specification’s authorization flow: dynamic client registration, a pre-registered OAuth client, or a credential the platform supplies. The operator chooses one per server in the connectors catalog.

Client capabilities the host advertises

Section titled “Client capabilities the host advertises”

On every initialize, the host sends exactly these capabilities:

{
  "capabilities": {
    "tasks": {
      "requests": { "tools": { "call": {} } },
      "cancel": {}
    },
    "extensions": {
      "ai.nimblebrain/host-resources": {
        "read": { "enabled": true, "range": false, "maxSize": 10485760 },
        "list": { "enabled": true },
        "write": { "enabled": false },
        "schemes": ["files"]
      }
    }
  }
}

tasks: the host runs long-running tools as tasks and can cancel them. If a tool declares execution.taskSupport as optional or required, the host calls it task-augmented, receives a CreateTaskResult right away, and waits for the result on the task stream. When the run that made the call is cancelled, the host sends tasks/cancel. A task-augmented call is never retried after a transport failure, because the task’s state lives on the server and a retry would start the work twice. The host does not claim tasks.list, because it never lists tasks.

extensions: the vendor capabilities described on NimbleBrain and MCP. Read this block before calling an extension method.

Client capabilities the host does not advertise

Section titled “Client capabilities the host does not advertise”
Capability Why not
sampling The specification is deprecating it. A server that needs a model should call one itself.
roots The specification is deprecating it. A server reaches workspace files through host-resources instead.
elicitation Elicitation asks the client to put a question to a person and wait for the answer. An agent here often runs with nobody present, such as a scheduled automation or a webhook-triggered run. Even when someone is present, they are at the far end of an event stream, not waiting behind the tool call. The host cannot honor “block until someone answers”, so it does not claim to. A tool that needs input should return that as its result, and the agent will ask.

What the host reads from your server

Section titled “What the host reads from your server”
Server feature How the host uses it
tools Offered to the agent, filtered by workspace, feature flags, and the caller’s role. annotations (readOnlyHint, destructiveHint, and the rest) and outputSchema are forwarded as declared. Annotations are the server’s claims about itself, so the host never uses them to relax a check.
_meta.ui.resourceUri on a tool Mounts the tool’s MCP Apps view inline with its result.
_meta.ui.visibility on a tool Absent means ["model", "app"]. A tool without "model" is left out of every tool list an agent sees, in chat and at the MCP endpoint. A tool without "app" is refused when an app’s view calls it, and a view can only call tools on its own server.
resources Read when the host needs them: ui:// views when they mount, resources at reserved URIs, and the server’s notification outbox. The host reads on its own schedule. If the server advertises resources.subscribe, the host subscribes to the outbox and treats notifications/resources/updated as a hint to read it sooner, not as a replacement for reading.
instructions in the initialize result Composed into the agent’s prompt for the workspace. See custom instructions.

As an MCP Apps host

Section titled “As an MCP Apps host”

The host implements the MCP Apps specification: ui/initialize, tool input and result delivery, ui/message, ui/open-link, ui/update-model-context, ui/download-file, ui/request-display-mode, size changes, and tools/call and resources/read proxied to the app’s own server. The MCP App Bridge reference lists every message and its exact behavior, including the three ai.nimblebrain/* bridge extensions.

As a server to external clients

Section titled “As a server to external clients”

Each workspace is its own MCP server at /mcp/<workspaceId>, over Streamable HTTP with OAuth 2.0 protected-resource discovery. It advertises:

Capability Notes
tools The workspace’s tools and the caller’s identity tools, with the same filtering the agent gets
resources Resources from the workspace’s servers
tasks tasks/cancel and task-augmented tools/call, so long-running tools stay long-running through the endpoint

It does not serve the optional GET server-to-client stream, so progress and task updates arrive on the POST that started them. The MCP endpoint reference covers authentication, sessions, and the audience rule.

Section titled “Related”