TL;DR:
In June, SmythOS introduced the launch of the open-source Smyth Runtime Environment (SRE), faster AI model updates, new deployment embodiment views, mobile user guidance, and expanded automated testing support. These changes make it easier to build, deploy, and test your agents more efficiently.
Everything listed below is already live across all SmythOS accounts.
Smyth Runtime Environment (SRE) Now Open Source
SRE, the core runtime that powers agent execution in SmythOS, is now available as an open-source project on GitHub.
SRE provides a unified API across all infrastructure providers, LLMs, storage, cache, vector databases, so you can build agent logic once and deploy it anywhere without vendor lock-in. It simplifies development, improves portability, and supports advanced security with built-in ACL and credential management.
SRE is for teams that need to move quickly in development while staying production-ready from day one.
Use the CLI to scaffold a project:
Agent Development and Deployment Enhancements
Hot-load AI models without redeploying
Swap in updated model files instantly. No redeployment required. Speeds up testing and iteration.
Embodiment management added to deployments
You can now view and manage linked embodiments directly inside the deployment screen. No more switching views to make changes.
Project scaffolding via CLI
Use the SmythOS CLI to quickly generate properly configured agent projects with just a few commands.
User Interface and Experience Improvements
Navigation simplified with “Build with AI”
We’re testing a more focused interface that replaces “Go to Dashboard” with “Build with AI” as your entry point.
Mobile usage guidance added
If you open the Builder on a phone or tablet, a one-time message explains limitations so you know what to expect.
Consistent dashboard layout
“Getting Started” cards now align with dashboard spacing and styling for a more polished and consistent UI.
Chat interface decluttered
Removed the attach icon from agent and test chats (except Weaver) for a cleaner experience.
Improved model list readability
Long model names and tags now wrap neatly, making them easier to scan and work with.
Automated Testing and QA Enhancements
Expanded data-qa locator coverage
New locators have been added to key areas like Home, Vault, My Plan, and Agent Settings. This gives QA teams better targeting and less brittle test automation.
Plan purchase test routes restored
We brought back test routes used for verifying plan purchases, improving support for end-to-end testing.
Feedback, Errors, and Workflow Guidance
Chat log export now shows clear errors
Invalid date entries now display error messages instead of failing silently.
Account deletion flow now adapts to user roles
The deletion modal now includes different instructions based on whether you’re a team member or account owner.
System Stability and Fixes
MCP embodiment connection bugs fixed
Resolved issues with browser and IDE connections to MCP. Connectivity is now stable and reliable.
Agent Weaver first response loads instantly
You’ll see the first response in a conversation without delay or extra clicks.
LLM output returns clean text by default
LLM responses now use plain text formatting to simplify downstream integrations and parsing.
Billing limit bugs resolved
Issues with setting and editing billing limits have been fixed for more accurate usage controls.
Claude model JSON parsing fixed
Fixed issues with malformed or incomplete responses from Claude models that were disrupting workflows.
UI cleanup across sidebars and forms
Removed unnecessary elements, fixed typos, and improved layout clarity throughout the interface.
Usage Monitoring and Notifications
Usage alert emails improved
Threshold alerts (80% and 100%) now include a BCC to an internal address so your team can monitor usage without missing a beat.
New SmythOS Documentation Portal
We launched a new, open-source documentation designed to support you at every stage of your workflow.
What’s included:
- Step-by-step guides for building and deploying agents
- Explanations of key platform components
- Examples for automation, workflows, and SDK usage
You can search, bookmark, and contribute to the docs through the open repo.
Where to Start
- Use
Settings → Models to hot-load new AI models - Visit
Deployments to manage embodiments in-place - Install the SDK with
npm install @smythos/sdk - Scaffold a project with the CLI using
sre create - Explore the runtime: https://github.com/SmythOS/sre
- Visit the new documentation for tutorials and examples
