feat: Add VideoCaptioner CLI — AI video captioning with styled subtitles

[WEIFENG2333]

Summary

Adds cli-anything-videocaptioner, an agent-native CLI harness for VideoCaptioner — AI-powered video captioning tool (13.7k+ ⭐, PyPI).

Full pipeline: speech transcription → subtitle optimization → translation (38 languages) → video synthesis with styled subtitles. Free ASR and translation included, zero API key needed for basic usage.

Command Groups

Group	Commands	Description
transcribe	4 ASR engines	Speech → subtitles (bijian/jianying free, whisper-api, whisper-cpp)
subtitle	3 translators	Optimize + translate subtitles (LLM, Bing free, Google free)
synthesize	5 style presets	Burn styled subtitles into video (ASS outline + rounded background)
process	full pipeline	One command: transcribe → optimize → translate → synthesize
styles	—	List all subtitle style presets with parameters
config	show, set	Manage configuration
download	—	Download online video (YouTube, Bilibili, etc.)
session	status	Version and state info

Architecture

videocaptioner/agent-harness/
├── VIDEOCAPTIONER.md
├── setup.py
└── cli_anything/videocaptioner/
    ├── videocaptioner_cli.py       # Click CLI (9 commands + REPL)
    ├── core/                        # transcribe, subtitle, synthesize, pipeline
    ├── utils/vc_backend.py          # subprocess wrapper for videocaptioner CLI
    ├── skills/SKILL.md
    └── tests/                       # 14 unit + 12 e2e (26/26 passing)

Install

pip install videocaptioner          # the underlying tool
pip install -e videocaptioner/agent-harness   # the CLI-Anything harness

Type of Change

• New Software CLI — adds a CLI harness for a new application

For New Software CLIs

• VIDEOCAPTIONER.md SOP document exists at videocaptioner/agent-harness/VIDEOCAPTIONER.md
• SKILL.md exists inside the Python package
• Unit tests at tests/test_core.py are present and pass without backend (14 tests)
• E2E tests at tests/test_full_e2e.py are present (12 tests)
• README.md inside the package
• registry.json includes an entry for the new software
• README.md updated with new software entry (table + directory tree)
• repl_skin.py in utils/ is a copy from the plugin

General Checklist

• Code follows existing patterns and conventions
• --json flag is supported on all commands
• Commit messages follow conventional format
• Tested locally — 26/26 tests passing, real transcription/translation/synthesis verified

[yuh-yang]

Thanks! The test results and examples look great.