update: add podcast-creator skill

[mvanhorn]

Summary

Adds a podcast-creator skill that converts text scripts into podcast episodes using MiniMax TTS and Music APIs.

What it does

• Takes a text script (plain text, Markdown, or structured JSON with chapters)
• Generates narration via MiniMax TTS API (speech-2.8-hd) with configurable voice selection
• Generates intro/outro music via MiniMax Music API (music-2.5+)
• Assembles everything with ffmpeg into a final podcast mp3 with crossfading and ID3 tags

Why

The TTS and Music APIs are only used inside frontend-dev for web asset generation. This skill surfaces them for audio content creation. The existing minimax_tts.py and minimax_music.py scripts serve as the foundation. The new podcast_create.py orchestrator handles chapter splitting, voice assignment, and ffmpeg assembly.

Structure

skills/podcast-creator/
  SKILL.md                          # Skill definition with 6-step workflow
  scripts/
    podcast_create.py               # Audio assembler (crossfade + concat + ID3)
    minimax_tts.py                  # TTS script (copied from frontend-dev)
    minimax_music.py                # Music script (copied from frontend-dev)
  references/
    requirements.txt                # Python deps
    script-format.md                # Input format documentation

Follows the same pattern as gif-sticker-maker: SKILL.md with mandatory workflow steps, scripts dir with self-contained Python CLIs, references dir with supplementary docs.

Test plan

• Verify SKILL.md frontmatter parses correctly
• Run python3 -m py_compile scripts/podcast_create.py (passes)
• Test with MINIMAX_API_KEY: generate narration, music, and assemble

This contribution was developed with AI assistance (Claude Code).

[wu335230960]

恩啊