Skip to content

Guide · Agent skill

How to use the VideoCaptioner Subtitles agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The VideoCaptioner Subtitles skill (cli-anything-videocaptioner) lets an AI agent drive a command-line video captioning tool for you. Once it is installed, you can ask the agent to transcribe speech in a video, clean up and translate the subtitles, and either attach them to the video or burn them into the picture. This guide explains what the skill does, how to install it, how to phrase requests, and what to check before you rely on it.

VideoCaptioner SubtitlesHKUDS/CLI-Anything · Apache-2.0

Quick start

  1. 1Download the skill. Download the skill ZIP from this page. It contains SKILL.md.
  2. 2Unzip into a skills folder. Place the folder in ~/.claude/skills/ for personal use, or in .claude/skills/ inside a project. Alternatively, upload it under Settings → Capabilities → Skills in the Claude apps.
  3. 3Check your Python version. Use Python 3.10 to 3.12. Python 3.12 is the recommended choice in the source.
  4. 4Install the packages and FFmpeg. Run pip install videocaptioner and pip install cli-anything-videocaptioner, and make sure FFmpeg is installed.
  5. 5Ask for a transcript. Request a transcription of your video. The default bijian option needs no setup.
  6. 6Translate or burn subtitles. Ask for translation into a target language, or for a hard-burned video once subtitles are ready.
  7. 7Review before final export. For hard burns, ask the agent to run the review step against your approved script and check the preview frame.

What the skill lets the agent do

The skill gives the agent access to the cli-anything-videocaptioner command-line tool, which wraps the VideoCaptioner backend. The source describes a pipeline: transcribe speech to subtitles, optimize the subtitles, translate them, and burn them into the video.

The agent can also render a single preview frame so you can check subtitle wording against an approved script before the final export.

  • Transcribe a video into subtitle files (srt, ass, txt or json)
  • Optimize and translate subtitles into one of 38 target languages
  • Attach subtitles to the video (soft) or permanently burn them in (hard)
  • Check subtitle wording against a script and render a one-frame preview
  • Download an online video with the download command

Install the skill from this page

Download the skill ZIP from this page. The skill package contains a single file, SKILL.md. You can install it in one of two ways.

Option A: unzip into your personal skills folder ~/.claude/skills/, or into a project's .claude/skills/ folder if you want it limited to one project.

Option B: in the Claude apps, open Settings, go to Capabilities, and upload the skill under Skills.

  • Personal install: ~/.claude/skills/
  • Project install: .claude/skills/ inside your project folder
  • Claude apps: Settings → Capabilities → Skills

Install the required Python packages and FFmpeg

The skill needs Python 3.10 to 3.12. The videocaptioner 1.4.1 package requires a version from 3.10 up to but not including 3.13, so Python 3.12 is the recommended choice for matrix runs. FFmpeg is required for video synthesis, meaning burning subtitles into a video.

Install the two Python packages with pip. The source lists videocaptioner first, then the CLI package.

  • Python 3.10, 3.11 or 3.12
  • FFmpeg installed and available for video synthesis
  • videocaptioner Python package
  • cli-anything-videocaptioner Python package
  • Whisper API key only if you choose the whisper-api transcription option
pip install videocaptioner pip install cli-anything-videocaptioner

Trigger the skill with natural requests

You do not need to type commands yourself. Describe the result you want and mention the video or subtitle file. The agent chooses the matching command. The examples below show the kind of request that maps to each part of the tool.

If you want to see the commands yourself, the source lists --help and an interactive REPL mode that starts when you run the CLI without a subcommand.

  • "Transcribe interview.mp4 into English subtitles."
  • "Translate subs.srt into Japanese and save the result."
  • "Make a hard-burned English version of promo.mp4 and check the text against approved_script.txt first."
  • "Render a preview frame at 5 seconds so I can check the wording."

Know which transcription option to use

The bijian option is the default and is free, covering Chinese and English with no setup. The whisper-api option supports all languages but requires a Whisper API key. The source also lists jianying and whisper-cpp as transcription choices without describing their requirements, so check the tool's help output before using them.

For translation, the Bing and Google translators are described as free. The llm translator is also listed, but the source does not say what it needs.

  • bijian: free, Chinese and English, no setup
  • whisper-api: all languages, needs a Whisper API key
  • bing and google translators: free
  • Target languages use BCP 47 codes such as en, ja, ko, fr, de and zh-Hans
cli-anything-videocaptioner transcribe video.mp4 --asr bijian cli-anything-videocaptioner subtitle input.srt --translator bing --target-language en

Review subtitles before a hard burn

Burning subtitles into the picture is permanent in the output file, so the source offers a review step first. The review command detects drift between your subtitles and an approved script, and can render one frame to check the wording.

The synthesize command can also run the same check with --review-script. The source advises lowering --max-script-diff-ratio when copy accuracy matters more than resilience.

  • Use review to compare subtitles against a script
  • Use --preview-at to pick the timecode for the preview frame
  • Advanced style flags depend on the installed backend; run styles to see what it supports
cli-anything-videocaptioner review subtitles.srt \ --script approved_script.txt \ --preview-video video.mp4 \ --preview-output review_5s.png

Safety notes: scripts and network use

This skill runs scripts and uses the network. The optional whisper-api transcription option and the online translators send content to external services. The source does not say what those services retain, so do not send confidential or sensitive recordings through them without checking their policies first.

The bijian transcription and Bing or Google translators are the options the source describes as free. Note that free options still involve the network only where the tool requires it; the source does not give further detail, so review the tool's behaviour yourself if this matters.

  • Do not use whisper-api or online translators for confidential recordings unless you have checked their data policies
  • Confirm the output path with -o before overwriting files
  • Use --json when an agent needs machine-readable output

What people use it for

  • Turning a lecture or interview recording into an SRT file for notes or captions
  • Translating a short promotional video's subtitles into several languages with the free Bing translator
  • Burning English subtitles into a short clip for social posting
  • Checking subtitle wording against an approved script before a final hard-burn export
  • Downloading an online video and then generating subtitles for it

Frequently asked questions

Is the VideoCaptioner skill free?

The source describes the bijian transcription option and the Bing and Google translators as free. The whisper-api option requires a Whisper API key, and the source does not give prices for it. The skill is published under the Apache-2.0 licence.

What do I need to install cli-anything-videocaptioner?

You need Python 3.10 to 3.12, the videocaptioner Python package, the cli-anything-videocaptioner package, and FFmpeg for video synthesis. A Whisper API key is only needed if you use the whisper-api transcription option.

How do I install the skill in Claude?

Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder. You can also upload it in the Claude apps under Settings → Capabilities → Skills.

Why does installation fail on my Python version?

The videocaptioner 1.4.1 package requires Python greater than or equal to 3.10 and less than 3.13. Python 3.13 or older than 3.10 will not match that range, so install Python 3.12 and try again.

Can the skill translate subtitles into any language?

The source says 38 target languages are supported, using BCP 47 codes such as en, ja, ko, fr, de, es and zh-Hans. It does not list every language, so check the codes you need against the tool.

Does the skill send my videos to the internet?

The skill uses the network. The optional whisper-api and the online translators send content to external services, and the source does not say what they retain. Avoid them for sensitive or confidential material unless you have checked their policies.

Can I control the subtitle style?

The source says advanced style flags depend on the installed backend version. Run the styles command to see whether preset styling is supported, and treat compatibility flags as non-stable unless the backend confirms them.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use this skill if you need quick subtitles, translations or hard-burned captions for short videos, lectures or promotional clips, and you can install Python 3.10 to 3.12 and FFmpeg. Pick another tool if your recordings are confidential and you cannot accept sending them to external services, or if you need guaranteed control over subtitle styling.

Download the skill