Agent Skills: Parallel PDF to Audio Converter

Parallel PDF to Audio Converter

Convert multiple PDF chapters to MP3 audio in parallel by delegating to /pdf2audio-minimax.

Trigger Phrases

This skill triggers on natural language requests:

| User Says | Action | |-----------|--------| | "generate mp3 for the story of 我的故事" | Convert all chapters | | "为故事我的故事生成mp3" | Convert all chapters | | "为故事我的故事第[01, 03-04]章生成mp3" | Convert specified chapters | | "generate mp3 for story of 我的故事, chapter [01-05]" | Convert specified chapters |

Input Format

Direct command:

/pdf2audio-parallel "<story_dir>/chapters/<story_name>-[chapter_pattern].pdf" [voice_id]

Examples:

• /pdf2audio-parallel "重写时间的源代码/chapters/重写时间的源代码-[01-05].pdf"
• /pdf2audio-parallel "我的故事/chapters/我的故事-[01, 03-04].pdf" "Chinese (Mandarin)_Gentleman"

Workflow

1. Parse Input Parameters

From natural language: Extract story name, then discover chapters:

# Find story directory
ls -d <STORY_NAME>/

# List available PDF chapters
ls <STORY_NAME>/chapters/<STORY_NAME>-*.pdf

From direct command: Extract from path pattern:

• story_dir: Story directory name (e.g., 重写时间的源代码)
• story_name: Story name from filename pattern (usually same as directory)
• chapter_pattern: Chapter numbers (e.g., 01-05 or [01, 03-04])
• voice_id: Voice ID for TTS (optional, passed to pdf2audio-minimax)

2. Generate Chapter List

Parse the chapter pattern and expand into individual files.

Supported Patterns:

| Pattern | Expands To | |---------|------------| | [01-05] | 01, 02, 03, 04, 05 | | [03] | 03 | | [01, 03-04] | 01, 03, 04 | | [01-02, 05-07] | 01, 02, 05, 06, 07 | | [01, 03, 05] | 01, 03, 05 |

Example: 我的故事/chapters/我的故事-[01, 03-04].pdf expands to:

• 我的故事/chapters/我的故事-01.pdf
• 我的故事/chapters/我的故事-03.pdf
• 我的故事/chapters/我的故事-04.pdf

3. Launch Parallel Agents

Use Task tool to spawn N agents simultaneously, each calling /pdf2audio-minimax for one PDF.

Agent Prompt Template:

Convert this PDF to MP3 using the pdf2audio-minimax skill:

/pdf2audio-minimax <story_dir>/chapters/<story_name>-XX.pdf [voice_id]

Example:
/pdf2audio-minimax 我的故事/chapters/我的故事-01.pdf Chinese (Mandarin)_Gentleman

IMPORTANT:

• Each agent delegates to /pdf2audio-minimax which handles:
  • Reading PDF content
  • Extracting chapter title from content
  • Voice selection (auto or specified)
  • Audio conversion via MiniMax
  • File naming: <故事名>_<章节号>_<章节标题>.mp3
• This skill only handles parallel orchestration

4. Collect Results

Gather outputs from all agents and report:

• Successfully converted files with final paths
• Any errors encountered

Quick Examples

Example 1: Simple Range

User Input:

/pdf2audio-parallel "我的故事/chapters/我的故事-[01-03].pdf"

Parallel Agent Tasks:

1. Agent 1: /pdf2audio-minimax 我的故事/chapters/我的故事-01.pdf → 我的故事_01_<标题>.mp3
2. Agent 2: /pdf2audio-minimax 我的故事/chapters/我的故事-02.pdf → 我的故事_02_<标题>.mp3
3. Agent 3: /pdf2audio-minimax 我的故事/chapters/我的故事-03.pdf → 我的故事_03_<标题>.mp3

Example 2: With Voice ID

User Input:

/pdf2audio-parallel "星际迷航/chapters/星际迷航-[01, 03-04].pdf" "Chinese (Mandarin)_Gentleman"

Parallel Agent Tasks:

1. Agent 1: /pdf2audio-minimax 星际迷航/chapters/星际迷航-01.pdf Chinese (Mandarin)_Gentleman
2. Agent 2: /pdf2audio-minimax 星际迷航/chapters/星际迷航-03.pdf Chinese (Mandarin)_Gentleman
3. Agent 3: /pdf2audio-minimax 星际迷航/chapters/星际迷航-04.pdf Chinese (Mandarin)_Gentleman

Notes

• Voice selection and file naming are handled by pdf2audio-minimax
• See pdf2audio-minimax skill for voice options and naming convention
• Output files are saved to <story_dir>/audiobook/ directory