SmythOS Weekly Update: Locator Requests, Gemini 3 Flash Model, Claude v4 (January 5 to January 11, 2026)

SmythOS Weekly Update

This week focused on improving day-to-day reliability and scalability when building, testing, and operating agents. This release expands model choice, strengthens authentication across embodiments, and reduces friction in QA, debugging, and configuration workflows through targeted UI and platform improvements.

New Features

Locator Requests: Expanded QA Attributes for Data Configuration

We added new data-qa attributes across key datasource configuration fields, including datasource name, chunk size, chunk overlap, metadata, and provider selection. These updates make UI elements easier to identify and validate during automated testing and QA workflows.

How this helps you
• Improves reliability of automated UI tests across data ingestion workflows
• Reduces brittleness in selector-based testing as the UI evolves
• Makes QA validation faster and more consistent for complex data setups

Gemini 3 Flash Model Support in GenAI LLM Components

SmythOS now supports the Gemini 3 Flash model across GenAI LLM components. This enables fast, cost-efficient inference with support for text, images, PDFs, audio, video, web search, reasoning, and function calling within existing GenAI workflows.

How this helps you
• Enables faster prototyping and real-time, multi-modal agent experiences
• Provides a high-throughput, cost-efficient option for large-scale inference
• Expands access to the latest Google AI capabilities without workflow changes

Global Variables and Inputs for Location and Time Fields

City, Country, Region, and Timezone fields in the GenAI component now support Global Variables and Inputs, allowing values to be dynamically populated with a single click.

How this helps you
• Makes agents more dynamic and reusable across contexts
• Reduces manual configuration for location-aware workflows
• Speeds up setup and iteration

Authentication Method Support Across All SmythOS Embodiments

Bearer Token and OAuth authentication are now supported across previously unsupported embodiments, including Voice, ChatGPT, LLM, Form Preview, MCP, and Alexa Voice. This ensures consistent authentication, security, and session management across the platform.

How this helps you
• Enables authenticated agents across all SmythOS interfaces
• Ensures consistent security and session behavior regardless of embodiment
• Removes functional gaps when deploying agents beyond the builder.

Improvements

Analytics CSV Export with Agent and Space Context

Analytics CSV exports now include Agent and Space columns, providing clearer attribution and context when reviewing activity outside the SmythOS UI.

How this helps you
• Makes it easier to analyze usage by specific agents and workspaces
• Improves reporting accuracy for audits and stakeholder reviews
• Reduces manual cleanup after exporting data

Data Pools Modal Design Consistency

Cancel buttons have been removed from Data Pools modals in favor of a single top-right close action, aligning these modals with SmythOS design standards.

How this helps you
• Creates a more consistent and predictable modal experience
• Reduces visual clutter in frequently used data workflows
• Reinforces platform-wide UI patterns

Additional QA Attributes for Locator Requests Components

Expanded data-qa selectors were added to provider selection, embedding model selection, and file upload inputs within Locator Requests.

How this helps you
• Improves reliability of automated UI testing
• Reduces selector breakage as the UI evolves
• Speeds up QA validation and regression testing

Semantic CSS Classes for Chat Controls

Chat send and stop buttons now use standardized semantic CSS classes to match Chatbot embodiments, without changing appearance or behavior.

How this helps you
• Enables consistent automation across Chat and Chatbot features
• Simplifies QA test maintenance
• Improves long-term reliability of test suites

OAuth Scopes Tooltip Clarification in Vault

A tooltip was added to the OAuth Scopes field clarifying that scopes must be space-separated and providing a clear example.

How this helps you
• Reduces OAuth configuration errors
• Improves onboarding for new integrations
• Lowers support requests related to setup issues

ForEach Component Log Readability Improvements

ForEach component logs now use clearer structure, indentation, and visual hierarchy to improve readability during workflow debugging.

How this helps you
• Speeds up debugging of iterative workflows
• Makes iteration-specific issues easier to identify
• Reduces cognitive load when reviewing logs

Bug Fixes

Agent Skill Advanced Request Toggle Default State

The Advanced Request Parts toggle now defaults to off, preventing advanced settings from appearing unless explicitly enabled.

How this helps you
• Reduces confusion when opening component settings
• Prevents accidental configuration changes
• Keeps the default experience clean

Upload Datasource Modal Button Size Consistency

Button height now remains consistent when expanding or collapsing Advanced Configuration options in the Upload Datasource modal.

How this helps you
• Improves visual stability and polish
• Reinforces trust in UI consistency
• Aligns with design system standards

Correct Data Spaces Display in RAG Component Settings

The RAG component now correctly displays data spaces from the currently selected space.

How this helps you
• Prevents accidental use of incorrect data sources
• Makes agent configuration more predictable
• Reduces friction when switching spaces

Clearer Behavior When Chatbot Is Disabled

Chat and Chatbot behavior now properly reflects disabled chatbot settings instead of failing silently.

How this helps you
• Reduces confusion when chat functionality is unavailable
• Improves transparency in agent behavior
• Lowers support burden

Chat Thinking Indicator Clears After Stop

Thinking and progress indicators now clear correctly when generation is stopped.

How this helps you
• Eliminates misleading UI states
• Improves confidence that execution has stopped
• Creates a clearer chat experience

Component Sidebar Search Shows Empty Sections

Searching by section name now displays empty component sections with clear indicators.

How this helps you
• Improves discoverability for new users
• Makes available integrations more visible
• Reduces onboarding confusion

Chat Send Button Disabled During File Upload

The chat send button is now disabled while files are uploading.

How this helps you
• Prevents failed file attachments
• Improves reliability of chat workflows
• Reduces user frustration

Agent Dashboard Grid Layout for Legacy Accounts

Visual gaps in the agent grid for legacy accounts have been resolved.

How this helps you
• Restores clean dashboard layouts
• Improves usability for long-time users
• Prevents layout confusion

Ollama Custom Model Logo Display Fix

Broken logos for custom Ollama models in the models dropdown have been fixed.

How this helps you
• Restores visual clarity in model selection
• Improves UI polish
• Makes custom models easier to identify

API Call Component Missing Configuration Indicator

The red notification badge now appears correctly when required API keys are missing.

How this helps you
• Makes configuration issues immediately visible
• Prevents failed executions
• Speeds up troubleshooting

Builder Scroll Issue After Feedback Modal

The builder page no longer locks scrolling after closing the feedback modal.

How this helps you
• Restores smooth canvas navigation
• Prevents workflow disruption
• Improves editing experience

Claude v4 Streaming Errors in Classifier and LLM Assistant

Execution errors affecting Claude v4 family models have been resolved.

How this helps you
• Restores compatibility with newer Claude models
• Prevents blocked workflows
• Improves execution reliability

Tooltips for View-Only Role Restrictions

Missing tooltips explaining disabled fields for view-only roles have been added.

How this helps you
• Clarifies permission-based restrictions
• Reduces confusion for non-editing users
• Improves role-based UX

Tooltip Padding Cleanup in Admin Pages

Excess padding in admin tooltips on Roles and User Management pages has been corrected.

How this helps you
• Improves visual polish in admin interfaces
• Keeps tooltips compact and readable
• Aligns with overall UI standards

Prettify No Longer Breaks Note Component Grouping

The Prettify feature now preserves grouped Note components and their children.

How this helps you
• Protects intentional canvas organization
• Prevents unexpected layout changes
• Saves time re-grouping components

Team Invitation Acceptance Email Grammar Fix

Incorrect possessive text in team invitation emails has been corrected.

How this helps you
• Improves professionalism in onboarding emails
• Reduces confusion for new team members
• Builds trust during first interactions

Vault Key Management in Expanded Modals

The Add Key option is now available in expanded textarea modals.

How this helps you
• Restores full key management functionality
• Keeps workflows consistent across contexts
• Reduces unnecessary navigation

GenAI LLM Stop Sequence Tooltip Visibility

The Stop Sequence tooltip now displays fully and clearly.

How this helps you
• Makes configuration guidance readable
• Reduces GenAI setup errors
• Improves sidebar usability

Grok 4 Web Search Real-Time Data Fix

Web search now correctly returns real-time data when enabled for Grok 4.

How this helps you
• Restores trust in time-sensitive queries
• Enables current-event use cases
• Improves production reliability

Try These Updates

Try Gemini 3 Flash in a GenAI LLM component
Switch an existing agent to Gemini 3 Flash and compare response speed, cost efficiency, and multi-modal handling for real-time or high-volume workflows.

Use Global Variables for location and time fields
Populate City, Country, Region, or Timezone using Global Variables or Inputs to make agents more reusable and context-aware.

Update QA tests to use new data-qa attributes
Leverage the expanded QA selectors in locator and data ingestion flows to improve test stability and reduce brittle selectors.

Export analytics with Agent and Space context
Download a CSV from Analytics and use the new columns to analyze usage across agents and workspaces more easily.

Review authentication across embodiments
Enable OAuth or Bearer Token authentication for agents deployed in Chat, Voice, Form Preview, MCP, or Alexa to unlock authenticated use cases.For a deeper understanding and additional examples, refer to the official SmythOS documentation, which provides a more detailed exploration of these changes.To schedule a meeting or request a demo, click here