/ai-daily

name: ai-daily-news
description: Fetches AI news from smol.ai RSS and generates structured markdown with intelligent summarization and categorization. Optionally creates beautiful HTML webpages with Apple-style themes and shareable card images. Use when user asks about AI news, daily tech updates, or wants news organized by date or category.

AI Daily News

Fetches AI industry news from smol.ai, intelligently summarizes and categorizes using built-in Claude AI capabilities, outputs structured markdown, and optionally generates themed webpages and shareable card images.

Quick Start

# Yesterday's news
昨天AI资讯

# Specific date
2026-01-13的AI新闻

# By category
昨天的模型发布相关资讯

# Generate webpage
昨天AI资讯，生成网页

# Generate shareable card image
昨天AI资讯，生成分享图片
生成日报卡片图片

Supported Query Types

| Type | Examples | Description | |------|----------|-------------| | **相对日期** | "昨天AI资讯" "前天的新闻" "今天的AI动态" | Yesterday, day before, today | | **绝对日期** | "2026-01-13的新闻" | YYYY-MM-DD format | | **分类筛选** | "模型相关资讯" "产品动态" | Filter by category | | **网页生成** | "生成网页" "制作HTML页面" | Optional webpage generation | | **图片生成** | "生成图片" "生成分享卡片" "制作日报卡片" | Generate shareable card image |

---

Workflow

Copy this checklist to track progress:

Progress:
- [ ] Step 1: Parse date from user request
- [ ] Step 2: Fetch RSS from smol.ai
- [ ] Step 3: Check if content exists for target date
- [ ] Step 4: Extract and analyze content
- [ ] Step 5: Generate structured markdown
- [ ] Step 6: Ask about webpage generation (if requested)
- [ ] Step 7: Generate shareable card image (if requested)

---

Step 1: Parse Date

Extract the target date from user request.

| User Input | Target Date | Calculation | |------------|-------------|-------------| | "昨天AI资讯" | Yesterday | today - 1 day | | "前天AI资讯" | Day before yesterday | today - 2 days | | "2026-01-13的新闻" | 2026-01-13 | Direct parse | | "今天的AI动态" | Today | Current date |

**Date format**: Always use `YYYY-MM-DD` format (e.g., `2026-01-13`)

---

Step 2: Fetch RSS

Run the fetch script to get RSS data:

python plugins/ai-daily/skills/ai-daily/scripts/fetch_news.py

This downloads and parses `https://news.smol.ai/rss.xml`, returning structured JSON.

**Available dates** can be checked with:

python plugins/ai-daily/skills/ai-daily/scripts/fetch_news.py --date-range

---

Step 3: Check Content

Verify if content exists for the target date.

When Content Exists

Continue to Step 4.

When Content NOT Found

Display a friendly message with available dates:

抱歉，2026-01-14 暂无资讯

可用日期范围: 2026-01-10 ~ 2026-01-13

建议:
- 查看 [2026-01-13](command:查看2026-01-13的资讯) 的资讯
- 查看 [2026-01-12](command:查看2026-01-12的资讯) 的资讯

**User experience principles**: 1. Clear problem statement 2. Show available alternatives 3. Provide clickable commands for quick access 4. Never leave user stuck with no options

---

Step 4: Extract and Analyze Content

Use built-in Claude AI capabilities to:

1. **Extract full content** from the RSS entry 2. **Generate summary** - 3-5 key takeaways 3. **Categorize** items by topic:

• Model Releases (模型发布)
• Product Updates (产品动态)
• Research Papers (研究论文)
• Tools & Frameworks (工具框架)
• Funding & M&A (融资并购)
• Industry Events (行业事件)

4. **Extract keywords** - Companies, products, technologies

**Prompt Template**:

Analyze this AI news content and organize it:

1. Generate 3-5 key takeaways (one sentence each)
2. Categorize items into: Model Releases, Product Updates, Research, Tools, Funding, Events
3. Extract 5-10 keywords

Original content:
{content}

---

Step 5: Generate Markdown

Output structured markdown following the format in [output-format.md](references/output-format.md).

**Key sections**:

• Title with date
• Core summary
• Categorized news items
• Keywords
• Footer with source info

**Example output**:

# AI Daily · 2026年1月13日

## 核心摘要

- Anthropic 发布 Cowork 统一 Agent 平台
- Google 开源 MedGemma 1.5 医疗多模态模型
- LangChain Agent Builder 正式发布

## 模型发布

### MedGemma 1.5
Google 发布 4B 参数医疗多模态模型...
[原文链接](https://news.smol.ai/issues/26-01-13-not-much/)

## 产品动态

...

## 关键词

#Anthropic #Google #MedGemma #LangChain

---
数据来源: smol.ai

---

Step 6: Webpage Generation (Optional)

**Only trigger when user explicitly says**:

• "生成网页"
• "制作HTML页面"
• "生成静态网站"

Ask User Preferences

是否需要生成精美的网页？

可选主题:
- [苹果风](command:使用苹果风主题) - 简洁专业，适合技术内容
- [深海蓝](command:使用深海蓝主题) - 商务风格，适合产品发布
- [秋日暖阳](command:使用秋日暖阳主题) - 温暖活力，适合社区动态

Theme Prompt Templates

See [html-themes.md](references/html-themes.md) for detailed prompt templates for each theme.

**Apple Style Theme** (Key Points):

Generate a clean, minimalist HTML page inspired by Apple's design:

**Design**:
- Pure black background (#000000)
- Subtle blue glow from bottom-right (#0A1929 → #1A3A52)
- Generous white space, content density ≤ 40%
- SF Pro Display for headings, SF Pro Text for body
- Smooth animations and hover effects

**Structure**:
- Header: Logo icon + Date badge
- Main: Summary card + Category sections
- Footer: Keywords + Copyright

**Colors**:
- Title: #FFFFFF
- Body: #E3F2FD
- Accent: #42A5F5
- Secondary: #B0BEC5

Save Webpage

Save to `docs/{date}.html`:

# Save webpage
cat > docs/2026-01-13.html << 'EOF'
{generated_html}
EOF

---

Step 7: Shareable Card Image Generation (Optional)

**Trigger when user explicitly requests**:

• "生成图片"
• "生成分享卡片"
• "制作日报卡片"
• "生成卡片图片"
• "生成分享图"

Image Generation Process

1. **Build condensed Markdown** for card display:

• Title and date
• Core summary (3-5 items)
• Top items per category (3 items each)
• Keywords

2. **Call Firefly Card API**:

• API: `POST https://fireflycard-api.302ai.cn/api/saveImg`
• Body contains `content` field with Markdown
• Returns binary image stream (`Content-Type: image/png`)

3. **Save and display result**:

• Save to `docs/images/{date}.png`
• Display