This week’s update focuses on the “under the hood” mechanics that make SmythOS a professional-grade development environment. We’ve prioritized the integrity of our telemetry data and the precision of our debugging tools, ensuring that when you build and test complex workflows, the feedback you receive is accurate, visual, and actionable.
Bug Fixes
Telemetry Log Data Cleanup We resolved an issue where “key and value” entries and “unknown” strings were appearing sporadically in the Telemetry & Logs view. These ghost values were not part of the actual execution flow but were cluttering span attributes and events.
- How this helps you: This restores absolute trust in your logs. By removing metadata noise, you can debug execution flows faster and rely on telemetry data for accurate performance analysis and auditing.
Chatbot Embodiment Debugger Visibility Fixed a bug in the Chatbot embodiment where the debugger failed to display Input/Output values on hover. While the standard builder debugger worked correctly, the chatbot version was withholding these critical data points after successful executions.
- How this helps you: This brings full transparency to the chatbot testing environment. You can now inspect exactly what data is passing through your workflow components in real-time, matching the power of the standard builder debugger.
Logic Component Execution Highlighting We corrected a UI oversight where Logic components remained “neutral” after execution, missing their status highlighting (green for success, red for failure) and the associated debug log button.
- How this helps you: You no longer have to guess if a logic branch executed correctly. Visual status highlighting provides immediate confirmation of your workflow’s path, while the restored log button allows for deep inspection of logic-gate decisions.
Google AI (Gemini) Custom Output Mapping Resolved a critical failure in the GenAI LLM component where Google AI models were failing to respect custom output mapping. Instead of returning structured JSON with custom keys (like ‘animal’ or ‘background’), the system was returning improperly formatted responses.
- How this helps you: This restores the ability to perform structured data extraction using Google’s models. Your downstream components that rely on specific JSON keys will now receive perfectly formatted data, preventing breaks in complex image analysis or text classification chains.
Try These Updates
- Verify Your Logic: Open a workflow with Logic components and run it in debug mode to see the restored status highlighting and log access.
- Inspect Your Chatbot: Launch a Chatbot embodiment with the debugger active and hover over components to see your Input/Output values in action.
- Test Structured Extraction: Use a Gemini model with custom output keys to confirm your data is mapping correctly to your structured JSON schema.
For a deeper understanding and additional examples, refer to the official SmythOS documentation. To see these debugging tools in action, click here.
SmythOS is securing Clawbot. Don’t get caught in a pinch, learn more about how SmythOS provides the unbreakable shell for your Clawbot agents.