/tldr

Extract and summarize a TLDR newsletter. Follow these steps precisely:

Step 1: Load TLDR Skill & Environment

• Read the TLDR skill at ${CLAUDE_PLUGIN_ROOT}/skills/tldr-scraper/SKILL.md.

Step 2: Determine Category & Language

The user's argument is: $ARGUMENTS

• If a category is specified (e.g., "ai", "dev"), use it to filter Gmail search.
• If no category, search for the most recent TLDR email and let the user pick, or process the most recent one.
• Determine the output language per the skill's Step 0 rules.

Step 3: Extract & Summarize

Follow the skill's extraction workflow (Steps 1-4): search Gmail → parse email → fetch articles sequentially → generate summaries.

Step 4: Save Locally

Save the assembled Markdown file to the user's workspace:

• Save to: {workspace}/outputs/tldr-{category}-news-YYYY-MM-DD.md
• Create the directory if it doesn't exist
• Present the local file path to the user

Step 5: Launch Subagents

After saving the Markdown file, launch two subagents in parallel using the Agent tool. This offloads context-heavy work to fresh context windows, improving output quality.

IMPORTANT: Send BOTH Agent tool calls in a single message so they run concurrently.

Before launching, prepare the following variables from the work done in Steps 1-4:

• markdown_path: the full path to the saved Markdown file
• view_online_url: the “View Online” URL extracted from the email
• newsletter_name: e.g., “TLDR AI - 2026-03-17”
• date_str: e.g., “2026-03-17”
• category: e.g., “ai”
• workspace: the workspace root path

Subagent A: notion-uploader (background)

Launch with run_in_background: true. Prompt:

Upload the TLDR newsletter to Notion.

- Markdown file: {markdown_path}
- Newsletter name: {newsletter_name}
- View Online URL: {view_online_url}

Subagent B: podcast-producer (foreground)

Launch with run_in_background: false. Prompt:

Generate a podcast from the TLDR newsletter summary.

- Markdown file: {markdown_path}
- Date: {date_str}
- Category: {category}
- Workspace: {workspace}
- Script output: {workspace}/outputs/tldr-{category}-podcast-{date_str}.txt
- Audio output: {workspace}/outputs/tldr-{category}-podcast-{date_str}.mp3
- TTS script: ${CLAUDE_PLUGIN_ROOT}/scripts/tts_minimax.py

After both agents complete

Present a summary of all outputs:

• Local Markdown file path
• Notion "AI资讯速递" page URL (from notion-uploader)
• Podcast script path and character count (from podcast-producer)
• Notion Podcasts page URL (from podcast-producer)
• Audio file path and size (from podcast-producer)