Videosays Agent Tools
AI-agent-friendly CLI and SKILL.md package for video transcription, video to text, speech to text, subtitle extraction, and transcript export.
Videosays turns supported video links or share text into clean transcript text, timestamped timelines, SRT subtitles, and VTT subtitles. It is built for humans using a terminal and for AI agents using a skill.
中文：Videosays 是面向 AI Agent 和命令行用户的视频转文字工具，支持把公开视频链接或分享文本转成纯文本、带时间轴文本、SRT 字幕和 VTT 字幕。
Supported platforms include Douyin, TikTok, Instagram Reels, X (Twitter), Xiaohongshu, Bilibili, YouTube, and Kuaishou. Availability can vary by source video accessibility, region, platform restrictions, and whether captions or transcribable audio are available.
支持平台包括抖音、TikTok、Instagram Reels、X（Twitter）、小红书、Bilibili、YouTube、快手。实际可用性会受视频访问权限、地区、平台限制、字幕或音频可获取性影响。
This repository contains:
packages/cli- npm CLI package, exposed as thevideosayscommand.skills/videosays- canonical agent skill that callsnpx videosays.SKILL.md- root copy for directories that require a repository-root skill file, such as SkillsLLM.
Quick Start
npx videosays login
npx videosays transcribe "https://www.tiktok.com/@creator/video/123456"
npx videosays transcribe "https://www.bilibili.com/video/BV1234567890" --format srt
The CLI stores your API key in ~/.videosays. You can also provide it through VIDEOSAYS_API_KEY.
首次使用会打开浏览器授权，CLI 会把 API key 保存到 ~/.videosays。也可以通过 VIDEOSAYS_API_KEY 环境变量提供。
Repeated videos reuse the same account's active task or completed result by default, without another transcription charge. Every submission carries an idempotency key so a network retry returns the same resource. Use --force-new only when a fresh, normally billed transcription is required; keep using status commands for later checks.
Agent Skill
Use Videosays from an AI agent by giving it this prompt:
Read https://videosays.com/SKILL.md, install the Videosays Skill if your environment supports skills, and help me transcribe this public video link.
Or install the skill from this repository:
npx skills add xwchris/videosays-agent-tools
Install through ClawHub or SkillUse:
clawhub install @wegofuture/videosays
skilluse repo add xwchris/videosays-agent-tools --path skills --branch main --default
skilluse install videosays --agent codex --global
skills/videosays/SKILL.md is the canonical skill entry for registry distribution. The root SKILL.md and the website copy at https://videosays.com/SKILL.md should stay byte-for-byte in sync with it.
skills/videosays/SKILL.md 是分发平台使用的 canonical skill。根目录 SKILL.md 和网站公开版本 https://videosays.com/SKILL.md 需要保持同步。
Output Formats
videosays transcribe "<video-link-or-share-text>" --format text # plain transcript, default
videosays transcribe "<video-link-or-share-text>" --format timeline # timestamped segments
videosays transcribe "<video-link-or-share-text>" --format srt # SRT subtitles
videosays transcribe "<video-link-or-share-text>" --format vtt # VTT subtitles
videosays status "<task-id>" --format srt
输出格式：
text: 纯文本，默认格式timeline: 带时间轴分段srt: SRT 字幕vtt: VTT 字幕
Skill Distribution
skills.sh / npx skills
skills.sh discovers skills from public GitHub repositories. This repository uses the standard skills/videosays/SKILL.md layout:
npx skills add xwchris/videosays-agent-tools
npx skills find videosays
ClawHub / OpenClaw
Install the published skill from the Wegofuture organization:
clawhub install @wegofuture/videosays
clawhub inspect @wegofuture/videosays
Publish updates after signing in to ClawHub with access to the wegofuture publisher:
clawhub login
clawhub skill publish skills/videosays --owner wegofuture
SkillUse
SkillUse can install skills from GitHub-backed repositories:
skilluse repo add xwchris/videosays-agent-tools --path skills --branch main --default
skilluse install videosays --agent codex --global
The --path skills flag is required because this repository keeps the canonical skill at skills/videosays/SKILL.md for registry compatibility. Omit --global if you want SkillUse to install into the current project's local agent skill directory instead.
Maintenance
Keep the canonical skill, root skill copy, and website copy in sync:
bash scripts/check-skill-sync.sh
VIDEOSAYS_PUBLIC_SKILL_PATH=/path/to/video2txt/packages/web/public/SKILL.md bash scripts/check-skill-sync.sh
SkillsLLM
SkillsLLM expects a SKILL.md file at the root of the submitted GitHub repository. Submit this repository URL and use the root SKILL.md copy:
https://github.com/xwchris/videosays-agent-tools
Commands
videosays login
videosays login --api-key <api-key>
videosays logout
videosays whoami
videosays transcribe <video-link-or-share-text>
videosays transcribe <video-link-or-share-text> --format text
videosays transcribe <video-link-or-share-text> --format timeline
videosays transcribe <video-link-or-share-text> --format srt
videosays transcribe <video-link-or-share-text> --format vtt
videosays status <taskId>
videosays status <taskId> --format srt
videosays batch <links.txt>
videosays batch status <batch-id>
videosays batch continue <batch-id>
videosays batch cancel <batch-id>
videosays balance
videosays history [limit]
videosays help
Batch progress checks use lightweight status responses. The CLI downloads complete Task results once at the end, and preserves completed work when a credit-limited batch is continued under the same Batch ID.
Links
- Website: https://videosays.com/?utm_source=videosays_skill&utm_medium=agent_skill&utm_campaign=videosays_agent_skill
- Public skill: https://videosays.com/SKILL.md
- API docs: https://videosays.com/docs?utm_source=videosays_skill&utm_medium=agent_skill&utm_campaign=videosays_agent_skill&utm_content=api_docs
- CLI package:
videosays
License
MIT