- Document worker and MinIO services in the production stack
- Drop --build from production compose command (pre-built image)
- Fix default URLs (8000, not 6003/8001) and dev log service name
- Add uploads/lib volumes to docker run example
- Note .env mount requirement and admin bootstrap defaults
- Restrict narration video export and download access to users with `is_admin` permission, removing the previous credit-based check.
- Update `project_workspace_routes` to ensure narration video tools are only enabled for admins.
- Refactor the todo board UI to use a more compact switch-based layout for transition pages and page numbering options.
- Update README image link.
Add optional TAVILY_BASE_URL environment variable and configuration option to enable custom Tavily API endpoints, such as proxies or private gateways. Updated core configuration, research service initialization, UI templates, and tests to handle the new base URL parameter. Also enhanced outline streaming with force_regenerate option for improved regeneration control.
Add OAuth, community, credits, startup, export, and narration workflows while splitting oversized route and PPT service modules into dedicated components with broader regression coverage.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Update default model from MiniMax-M2.1 to MiniMax-M2.7
- Fix API base URL from api.minimaxi.com to api.minimax.io
- Add temperature clamping (max 1.0) for MiniMax in get_provider_config()
- Add MiniMax/DeepSeek/Kimi env vars to reload_ai_config() (were missing)
- Update config service legacy migration to handle both old URLs/models
- Update UI template with correct MiniMax defaults and placeholders
- Add comprehensive test suite (21 unit + 3 integration tests)
- Update README/README_EN and .env.example with M2.7 docs
Co-authored-by: octopus <octopus@github.com>
- Added configuration options for using the OpenAI responses API and enabling reasoning in AIConfig.
- Implemented logic to handle responses API and reasoning in the landppt_api and related services.
- Updated the FastAPI application to support new configuration options and adjust request payloads accordingly.
- Enhanced the web interface to allow users to toggle responses API and reasoning settings.
- Added utility functions for extracting response data and usage metrics from OpenAI API responses.
- Introduced a lifespan context manager for managing application startup and shutdown processes.
- Created tests for the lifespan functionality to ensure proper handling of startup and shutdown events.
- Added project-related fields to PPTState for better context handling.
- Updated PPTOutlineGenerator to accept new parameters for project topic, scenario, requirements, audience, and style.
- Modified GraphNodes to incorporate project information in structure analysis and content generation.
- Comprehensive language optimization in prompt templates to ensure professional and structured output.
- Established strict guidelines to maintain content fidelity and prevent fabrication during PPT outline generation.
- Updated uv.toml for LandPPT with cache directory and python preference.
- Created comprehensive uv usage guide in Chinese.
- Added Windows setup script (setup.bat) for easy installation and environment setup.
- Implemented Python setup script (setup.py) for cross-platform environment configuration.
- Developed Unix/Linux/macOS setup script (setup.sh) for streamlined project initialization.
- Enhanced backend logic to skip already generated slides and differentiate between user-edited and AI-generated slides.
- Updated frontend to retain existing slides on page refresh and re-render them without clearing the container.
- Added handling for skipped slides to ensure proper display and progress tracking.
- Improved data persistence and continuity during the PPT generation process.
- Documented the implementation details and testing suggestions in the new feature documentation.
sligter