MeshAgent 0.41 makes four managed-agent and service workflows easier: inspecting managed-agent conversations, routing websocket processes through /messages, managing routes across SDKs, and controlling feed exports, create flows, and image deploy behavior.
Managed-agent support now includes thread listing, thread create/update/delete events, attachment names, and trusted sender-name handling. That gives clients enough structure to keep thread lists current, display attachment context, and replay chat history with fewer gaps between what the agent saw and what the UI shows.
For product surfaces like Studio, Powerboards, and custom room apps, the important change is consistency. A managed-agent conversation can be represented as a set of named threads with lifecycle events, attachments, participants, and sender identity that flow through the same APIs used by the SDKs.
/messages service pathMeshAgent 0.41 updates websocket process support to use the /messages path and adds explicit service authentication modes: jwt, iap, and none. The CLI also supports websocket-based process use, so developers can connect to a running process through the same message service shape used by hosted app routes.
That makes process-backed applications easier to reason about. A service can expose an interactive websocket endpoint, MeshAgent can route room messages to it, and the deployment can choose the authentication mode that fits the environment.
The TypeScript chatbot template was updated around this model too. It now builds and runs as a standalone Next.js app, deploys as a private websocket-enabled service, serves its message endpoint on /messages, and keeps chat messages pinned to the bottom of the UI.
The RouteSpec model introduced in 0.40 is now carried through the Python, TypeScript, Dart, and .NET SDKs. Clients can create, update, read, delete, and list routes using spec-based models, including both room and agent backends, while preserving compatibility with older route payloads.
This is useful for teams that manage app routes and agent routes from more than one language. A Python operator script, a TypeScript admin UI, a Dart app, and a .NET service can work with the same route concepts instead of each SDK interpreting route payloads slightly differently.
There are also schema cleanup notes for managed-agent SDK users. Toolkit metadata no longer includes the older thumbnailUrl and pricing fields, and some managed-agent invoke paths no longer carry callerContext. Code that depended on those fields should move to the current managed-agent schema.
Feed subscriptions now accept optional filename datetime formats across Python, TypeScript, Dart, .NET, and CLI surfaces. That gives teams a supported way to control how exported feed files are named when a subscription writes time-based output.
The meshagent create workflow also gets a few sharp edges removed. It now prints a cd hint after creating a project and blocks reusing an occupied nested folder, which helps avoid accidentally scaffolding into the wrong place.
Image deploys preserve Dockerfile default environment values and clear newly built images from the room cache. For operators, that means fewer surprises when a Dockerfile declares defaults and a freshly built image needs to be used by a room without stale cache behavior getting in the way.
Taken together, MeshAgent 0.41 makes managed-agent conversations easier to display and replay, websocket app routes clearer to operate, route APIs more consistent across SDKs, feed exports easier to name, and create/deploy workflows less error-prone.
Join our Discord community to keep up with MeshAgent releases, ask questions, and share feedback with the team.
Check out the MeshAgent documentation to start building today.
MeshAgent Studio, SDK, and Server give you everything to build, test, and deploy agentic applications, from development to production.
