MeshAgent 0.27 is a major platform release focused on first-class routing/domains, Codex-powered agents, and structured event tracking across MeshAgent Server, Studio, and SDKs.
Routes let you attach a domain to a room and point it at a published service port. Routes have full create, update, list, and delete support. You can also attach JSON annotations to routes for validation and routing behavior.
Routes allow you to deploy static sites or an agent-powered application on a shareable domain in minutes. This is particularly helpful for building agentic applications that expose core components of MeshAgent Rooms. The WebServer workflow builds on this by letting you serve static files or Python handlers from a WebServer config, run them locally, and deploy them with the route managed by the platform.
This keeps the path from local development to a public, managed endpoint simple, without stitching together separate hosting or proxy layers, and keeps routing explicit and auditable. As a simple example, you can now use built-in MeshAgent agents to build web-apps for you like a landing page that collects contact information and stores it in the in-room database.
Port and endpoint annotations control how incoming requests are validated and where they go. These annotations can route traffic to queues or processors and enforce validation before proxying. This allows you to enforce request safety and consistency at the platform level.
Codexagents are now first-class in MeshAgent. You can deploy Codex-powered agents directly as chatbots, task runners, or background workers, complete with approval policies, sandboxed execution, and skill directory support.
New capabilities:
1. meshagent codex chatbot with approval workflows for code execution
2. Sandbox policies to control file system access and network permissions
3. Skill directories for reusable Codex workflows
4. Multi-rules file support for complex agent configurations
5. Shell image selection and custom mounts
If you're building agents that write code, analyze data, or automate workflows, Codex support makes it seamless to add powerful execution capabilities while maintaining control and safety.
Agent actions (tools, web, exec, search, approvals, plans, diffs) are now normalized into structured events. These events are persisted in thread state and surfaced in Studio and SDKs for live status making it easier to debug, audit, and observe agent behavior over time
We've tightened security around how agents execute code and interact with containers:
1. Container exec now uses argv arrays instead of shell-joined strings, enabling safer execution
2. Secret-backed environment variables for runtime configuration are now supported
3. IAP (Identity-Aware Proxy) support for enterprise authentication flows
4. Approval policies for Codex so code execution requires human confirmation when needed
MeshAgent Studio gets major upgrades in 0.27:
1. Routes management UI: Create, edit, and delete routes with annotation editing and validation
2. Mailbox annotation editing: Configure email-based agent triggers directly in Studio
3. Resizable sidebar for better project navigation
4. Structured event rendering with approval action buttons
5. Live agent status indicators in chat
Dart and Python SDKs add full route APIs, mailbox annotation support, and structured event models. Python also adds meshagent‑codex and WebServer CLI workflows. Node.js/TypeScript and .NET have no functional changes.
Request annotation keys changed from meshagent.webhook.* to meshagent.request.*. CLI secret usage changed from meshagent secrets --secret-id ... to meshagent secret --id.
Join our Discord Server to stay up to date with MeshAgent releases, see full release notes, 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.
