MeshAgent 0.31 introduces first-class process agents through the new meshagent process workflow, improves how users manage active turns and threads in chat, and adds better queue tooling and service onboarding.
This release is focused on two practical improvements. First, it gives teams a new way to build one agent that can work across chat, email, queues, and tools from the same setup. Second, it makes live agent interactions easier to control, especially when a turn is still running and the user wants to stop, redirect, or better understand what the agent is doing.
MeshAgent 0.31 introduces first-class process agents through the new meshagent process workflow. Process agents let you build one agent that can work across chat, email, queues, and toolkits from the same running agent. This is useful when you want a single agent to answer users in chat, react to queued jobs, handle email workflows, and use room tools without splitting that behavior across separate agent setups.
In the CLI, this is exposed through the new meshagent process commands and repeatable --channel options.
This matters because many production agents do not live in just one interface. A support agent may answer questions in chat, pick up follow-up work from a queue, and send email updates. A coding or operations agent may chat with users while also exposing toolkit-based actions in the same room. MeshAgent 0.31 makes those workflows easier to build as one agent instead of stitching them together from separate pieces.
MeshAgent 0.31 improves how users steer active turns in both OpenAI and Anthropic-powered agents. In practice, this means users can cancel an in-progress response and redirect the agent with a new instruction in the same thread, instead of waiting for a stale tool chain to finish.
That matters when an agent is going down the wrong path, taking too long, or acting on outdated context. Instead of waiting for the turn to complete before correcting it, users can intervene sooner and keep the conversation moving.
This release also improves live chat and thread behavior across MeshAgent Studio, Flutter, and Powerboards. Waiting indicators are clearer, thread status is cleaner, duplicate-event issues have been fixed, thread parsing is richer, and process output can now be rendered with ANSI formatting.
For users, this makes active conversations easier to follow. It is clearer when an agent is working, easier to tell what state a thread is in, and easier to read structured output from process-based agents. These changes make long-running conversations feel more understandable and less noisy.
MeshAgent 0.31 continues the work of making services and MCP integrations easier to set up. The CLI now has clearer validation, service-template handling is improved, and public MCP servers now support automatic OAuth discovery.
The main benefit is less setup friction. If you are turning a service or MCP server into something usable in a room, the platform now does more to guide the setup and catch issues earlier.
The CLI now includes better room queue helpers for sending JSON payloads, queueing email messages, and receiving queue messages directly from a room.
This makes queue-based workflows easier to work with in practice. Teams can test real room behavior, push background jobs, and inspect queue traffic more directly from the CLI instead of building extra tooling just to exercise those paths.
MeshAgent 0.31 also updates the default CLI model selections to gpt-5.4 and gpt-5.4-codex.
This release improves room startup and provisioning diagnostics so failed jobs and terminated containers are detected and surfaced earlier.
That matters most when teams are deploying and operating services in active environments. Faster visibility into failures makes problems easier to diagnose and fix. MeshAgent 0.31 also adds Sentry webhook signature validation for service hooks and tightens build and publish workflows across SDK and app packages.
Powerboards gets better file and folder navigation, improved mobile layout polish, better context menu behavior, and stronger thread handling. MeshAgent Studio and shared Flutter chat components also get refinements to room and messaging workflows.
These changes are about making the product easier to use every day. As more teams rely on threaded and multi-surface agent workflows, these interface enhancements improve the user experience in MeshAgent.
Join our Discord Server to stay up to date with MeshAgent releases, ask questions, and share feedback with our 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.
