skybridge Releases

Major release adding infrastructure components and a new authentication example.

Mar 16, 2026

Feature release introducing middleware support and state generation for OAuth providers.

Mar 11, 2026

Updates the default model to `gpt-5-mini` with necessary configuration adjustments and introduces `SimulationExperimentEmpiricalValidator` for statistical validation, alongside Jupyter widgets and cost tracking utilities.

Feb 2, 2026

Adds automatic LLM model loading and API connection testing, while adding a workflow reminder for SelfHost configurations and fixing digital human voiceover parameter errors.

Jan 27, 2026

Introduces an 'Action Transfer' module for video generation, enables local workflow support for digital humans and video, and locks edge-tts version to ensure audio stability.

Jan 26, 2026

Deploys comprehensive pipelines for digital human voiceover and Audio-Visual video generation, expands TTS to multiple languages, and adds support for 48GB VRAM machines.

Jan 14, 2026

Enhances workflow options with Nano Banana image generation, adds ComfyUI API key configuration, and optimizes LLM structure parsing for better stability.

Dec 28, 2025

Adds a built-in FAQ sidebar, introduces new static_excerpt templates, and resolves TTS audio generation issues by locking edge-tts version to 7.2.1.

Dec 10, 2025

This release updates the default model to GPT-4.1-mini, which is reported to bring quality improvements, though users are advised to retest scenarios due to potential behavioral differences from the previous default of GPT-4o-mini.

Jul 31, 2025

This alpha release introduces the TinyTroupe research paper, parallel simulation capabilities, A/B testing support, and new verification tools, alongside significant improvements to agent generation and configuration management.

Jul 16, 2025

This release removes the TensorRT dependency from the project.

Feb 14, 2025

This SDK release introduces a new layout detection feature to improve document parsing capabilities.

Feb 14, 2025

This release focuses on improving document parsing accuracy with new layout and text detection features, alongside several bug fixes for strategy handling and testing.

Feb 11, 2025

This version adds layout detection functionality to the core megaparse library.

Feb 11, 2025

This alpha release introduces significant structural changes including deeper persona specifications, fragments for agent reuse, LLM-based logical propositions, and interventions for event-based modifications.

Jan 29, 2025

This release introduces a modular parser and formatter, type safety for strategy outputs, and text detection improvements, resolving multiple bugs related to engine configuration and testing.

Jan 16, 2025

Minimal release notes: "This includes the extended context size to 8192 so should provide 5 mins conversations on moshi-mlx."

Sep 22, 2024

Sep 20, 2024