Threads – Release Notes (v0.7)

Overview

Version 0.7 introduces full Gen-AI generation capabilities, including per-document prompt-driven content generation, provider configuration, OpenAI integration, and asynchronous multi-editor execution with UI-level busy indicators. This release establishes the foundation for a extendable, multi-provider LLM workflow while maintaining the existing document layout and editing model.

New Features in v0.7

Configurable AI Provider Framework

A new Configure… dialog provides a unified interface for system-level and provider-specific settings:

System Tab

• Edit the global System Prompt used for all generations.

OpenAI Tab

• API key (masked)
• Optional custom endpoint
• Preset model list:
  • gpt-4o-mini
  • gpt-4o
  • o3-mini
  • Other → reveals a custom model name field
• Support for model overrides when presets become outdated
• Automatic load/save of configuration in a user-specific app directory

Additional provider tabs (Gemini, DeepSeek, Ollama, Grok) are included as placeholders for future integrations.

Gen-AI Generation Workflow

Each document now supports prompt-based content generation:

• Users set a Prompt on any document.
• Selecting Edit → Generate triggers generation for the focused document.
• Generation uses:
  • Global System Prompt
  • Document Prompt

Per-Editor Async Generation

• Each document editor generates independently in parallel.
• Editors become temporarily read-only during generation.
• A semi-transparent overlay displays Generating… with an indeterminate progress bar.
• Sibling/Child creation buttons remain active.
• Generated text is written directly into the document’s Content field.

OpenAI Integration (First Provider Implementation)

A new abstraction layer encapsulates provider calls. Version 0.7 includes the first concrete backend:

OpenAI Chat Completion Backend

• Uses the official OpenAI SDK.
• Supports both default and custom endpoints.
• Converts internal document structures into Chat API messages.
• Returns full assistant text as document content.

This abstraction enables drop-in integration of additional providers in future versions.

Configuration Persistence

All provider and system settings are automatically stored as JSON in the user-local app directory:

• Loaded when opening the Configure dialog
• Saved on dialog close
• Ensures persistent environment across editor sessions

Known Issues

Mac:

• Currently configurations have difficulty saving on Mac, causing OpenAI API keys not able to be set. (Addressed in release v0.7.1)

Limitations

• No document deletion or rearrangement
• Markdown preview remains basic
• Only OpenAI is implemented; other providers are placeholders
• Generation does not yet stream partial output

References

• Basic Concept: Project and Document