Increase concurrency to 5 threads to reduce queue wait time

Add Bilibili demo video link to README

Fix Stage 1 reasoning box: always show expandable AI reasoning process

Add gitignore to exclude video materials folder

Fix short_description length to meet 60 char limit

Complete rebrand to AdGuardian: update all names, descriptions, and content to focus on advertising compliance

Rename to AdGuardian - more specific for advertising compliance

Update examples: 2 English + 1 Chinese with original content and proper language settings

Update README format following reference style with models and tags

Move reasoning box above content, fix clickable details, clean markdown format, set USA/English as default

Fix Qwen3-8B thinking issue, separate reasoning from content, improve HTML parsing for stages 2-3

Add openai and requests dependencies

Add four-tab interface for caching all stage results, enable Qwen3-8B thinking, and implement proper stage switching

Fix model configuration and implement real streaming output with proper reasoning_content parsing

Implement three-stage reasoning analysis with QwQ-32B and DeepSeek-R1 models, proper reasoning_content handling, and progressive context building

Implement AI-generated HTML with three progressive stages: analysis report with collapsible reasoning, annotated version, and revised text comparison

Implement three-stage progressive output: AI analysis with collapsible reasoning, annotated HTML version, and revised text comparison

Fix streaming analysis None value concatenation error and improve error handling

Fix AI models configuration: SiliconFlow (free) + Nebius (paid), convert hardcoded text to English

Replace dual analysis modes with single streaming analysis and AI model selection

Remove local word detection, implement real streaming output, increase font sizes and improve layout

Add country selection, language options, and streaming output

Implement detailed AI analysis with comprehensive reporting

Fix flagging parameter for Gradio compatibility

Implement real AI-powered content audit with Hanyu Xinxie style cards

Implement real AI-powered content audit with Hanyu Xinxie style cards

Fix flagging_mode parameter for Gradio compatibility

Fix short_description length

Add ContentGuardian - Content Audit Agent with Hanyu Xinxie Style

initial commit