Skip to content

Commit 83a3a8f

Browse files
committed
Add VideoDB skill listing
1 parent 9d67b44 commit 83a3a8f

3 files changed

Lines changed: 197 additions & 0 deletions

File tree

src/content/skills-zh/videodb.md

Lines changed: 83 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,83 @@
1+
---
2+
slug: videodb
3+
---
4+
5+
## 使用场景
6+
7+
- 从 YouTube、公开 URL、本地文件或 RTSP/实时视频流摄取视频。
8+
- 捕获桌面屏幕、麦克风和系统音频,用于实时感知工作流。
9+
- 为上传、生成、编辑或剪辑后的媒体返回可播放的 HLS 流链接。
10+
- 从视频和实时流中的视觉信息与语音信息生成实时上下文。
11+
- 为视频内容构建语音词索引、视觉索引、语义索引、关键词索引和时间索引。
12+
- 按转录文本、视觉场景、对象、动作、元数据或时间戳搜索精确片段。
13+
- 返回带时间戳、可播放证据链接和自动生成片段的搜索结果。
14+
- 生成干净的带时间戳转录文本和可样式化字幕。
15+
- 在服务端完成裁剪、合并、片段编译、字幕添加和媒体时间线合成。
16+
- 添加文字、图片、品牌元素、音频、背景音乐、旁白、配音和翻译叠加层。
17+
- 对编解码器、FPS、分辨率、码率、质量和宽高比进行转码与标准化。
18+
- 将视频重构为竖屏、方形、横屏等社交平台格式。
19+
- 生成 AI 媒体素材,包括图片、视频、音乐、音效和语音。
20+
- 监控 RTSP/实时视频流,并在事件发生时触发实时告警。
21+
- 录制桌面会话,总结发生的事情,并保留可搜索的会话记忆。
22+
23+
## 工作流
24+
25+
VideoDB 为 AI Agent 提供统一的服务端视频能力栈:
26+
27+
- **See(看见)**:摄取本地文件、公开 URL、YouTube 视频、RTSP/实时流,或包含屏幕、麦克风和系统音频的桌面会话。
28+
- **Understand(理解)**:生成实时上下文、转录文本、视觉索引、语义索引、带时间戳的记忆和可搜索证据。
29+
- **Act(行动)**:生成片段、字幕、叠加层、音频、配音、翻译、转码、重构、告警、导出结果和可播放流。
30+
31+
你的 Agent 只需要把视频交给 VideoDB,就能获得结构化上下文、可搜索片段、生成素材、告警、剪辑和 HLS 流链接,而无需在本地维护视频处理基础设施。
32+
33+
## 示例
34+
35+
```bash
36+
npx skills add video-db/skills
37+
```
38+
39+
然后向你的 Agent 提出请求:
40+
41+
```text
42+
上传这个 YouTube 视频,找到演讲者讨论定价的所有片段,并返回带时间戳的可播放剪辑。
43+
```
44+
45+
更多示例提示词:
46+
47+
```text
48+
上传 https://www.youtube.com/watch?v=MnrJzXM7a6o 并给我一个可分享的播放链接。
49+
```
50+
51+
```text
52+
截取 10s-30s 和 45s-60s 的片段,并把它们合成为一个视频。
53+
```
54+
55+
```text
56+
生成一段背景音乐,并把它添加到这个片段中。
57+
```
58+
59+
```text
60+
给原视频添加字幕,字幕使用黑底白字样式。
61+
```
62+
63+
```text
64+
找出所有出现手机特写或产品展示画面的场景。
65+
```
66+
67+
```text
68+
监控这个 RTSP 摄像头,并在有人进入房间时记录带时间戳的告警。
69+
```
70+
71+
```text
72+
捕获我的屏幕两分钟,并写一份说明我在做什么的报告,同时给出洞察和建议。
73+
```
74+
75+
## 注意事项
76+
77+
- 需要 Python 3.9+ 和 VideoDB API key。
78+
- 获取 $20 免费额度,无需信用卡:https://console.videodb.io
79+
- Claude Code 中的设置命令是 `/videodb setup`
80+
- 对于 Cursor、GitHub Copilot、Codex、opencode 和其他 Agent,请让 Agent 执行 “setup videodb”。
81+
- 也支持 Claude Code plugin 安装方式:`/plugin marketplace add video-db/skills``/plugin install videodb@videodb-skills`
82+
- 支持的宿主平台包括 macOS、Linux 和 Windows PowerShell;桌面捕获能力取决于具体宿主环境。
83+
- Skill 源码和后续更新保留在 VideoDB 官方 skills 仓库:https://github.com/video-db/skills

src/content/skills/videodb.md

Lines changed: 109 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,109 @@
1+
---
2+
name: videodb
3+
title: VideoDB - Perception, Memory, and Action for Video Agents
4+
description: The perception, memory, and action layer for AI agents. Ingest, understand, search, edit, transcribe, subtitle, capture, monitor, and stream video and audio through natural language.
5+
source: community
6+
author: VideoDB
7+
githubUrl: https://github.com/video-db/skills
8+
docsUrl: https://docs.videodb.io/
9+
category: data
10+
tags:
11+
- video
12+
- audio
13+
- perception
14+
- multimodal-ai
15+
- video-search
16+
- transcription
17+
- rtsp
18+
- hls
19+
- video-editing
20+
- ai-agents
21+
roles:
22+
- developer
23+
featured: false
24+
popular: false
25+
isOfficial: true
26+
installCommand: |
27+
npx skills add video-db/skills
28+
date: 2026-07-09
29+
---
30+
31+
## Use Cases
32+
33+
- Ingest videos from YouTube, public URLs, local files, or RTSP/live feeds.
34+
- Capture desktop screen, mic, and system audio for real-time perception workflows.
35+
- Return playable HLS stream links for uploaded, generated, edited, or clipped media.
36+
- Create real-time context from visual and spoken information in videos and live streams.
37+
- Build spoken-word, visual, semantic, keyword, and temporal indexes over video content.
38+
- Search exact moments by transcript, visual scene, object, action, metadata, or timestamp.
39+
- Return search results with timestamps, playable evidence links, and auto-generated clips.
40+
- Generate clean timestamped transcripts and styled subtitles.
41+
- Trim, merge, compile clips, add subtitles, and compose media timelines server-side.
42+
- Overlay text, images, branding, audio, background music, voiceover, dubbing, and translations.
43+
- Transcode and normalize codec, FPS, resolution, bitrate, quality, and aspect ratio.
44+
- Reframe videos for social formats such as vertical, square, and landscape.
45+
- Generate AI media assets including images, video, music, sound effects, and voiceovers.
46+
- Monitor RTSP/live feeds and trigger real-time alerts when events happen.
47+
- Record desktop sessions, summarize what happened, and preserve searchable session memory.
48+
49+
## Flow
50+
51+
VideoDB gives agents one consistent server-side video stack:
52+
53+
- **See**: ingest local files, public URLs, YouTube videos, RTSP/live feeds, or desktop sessions with screen, mic, and system audio.
54+
- **Understand**: create real-time context, transcripts, visual indexes, semantic indexes, timestamped memory, and searchable evidence.
55+
- **Act**: generate clips, subtitles, overlays, audio, dubbing, translations, transcodes, reframes, alerts, exports, and playable streams.
56+
57+
Your agent sends video in and gets back structured context, searchable moments, generated assets, alerts, clips, and HLS stream links without running local video infrastructure.
58+
59+
## Example
60+
61+
```bash
62+
npx skills add video-db/skills
63+
```
64+
65+
Then ask your agent:
66+
67+
```text
68+
Upload this YouTube video, find every moment where the speaker discusses pricing, and return playable clips with timestamps.
69+
```
70+
71+
More example prompts:
72+
73+
```text
74+
Upload https://www.youtube.com/watch?v=MnrJzXM7a6o and give me a shareable stream link.
75+
```
76+
77+
```text
78+
Take clips from 10s-30s and 45s-60s and compile them.
79+
```
80+
81+
```text
82+
Generate background music and add it to this clip.
83+
```
84+
85+
```text
86+
Add subtitles to the original video with white text on a black background.
87+
```
88+
89+
```text
90+
Find every scene showing a phone close-up or product on screen.
91+
```
92+
93+
```text
94+
Monitor this RTSP camera and log an alert with timestamp whenever a person enters the room.
95+
```
96+
97+
```text
98+
Capture my screen for two minutes and write a report of what I am doing with insights and suggestions.
99+
```
100+
101+
## Notes
102+
103+
- Requires Python 3.9+ and a VideoDB API key.
104+
- Get $20 free credits. No credit card needed: https://console.videodb.io
105+
- The Claude Code setup command is `/videodb setup`.
106+
- For Cursor, GitHub Copilot, Codex, opencode, and other agents, ask the agent to "setup videodb".
107+
- Claude Code plugin install is also supported with `/plugin marketplace add video-db/skills` and `/plugin install videodb@videodb-skills`.
108+
- Supported host platforms include macOS, Linux, and Windows PowerShell; desktop capture support depends on the host environment.
109+
- Skill source and updates stay in the official VideoDB skills repository: https://github.com/video-db/skills

src/i18n/skills-translations.ts

Lines changed: 5 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -321,6 +321,11 @@ export const skillsTranslations: Record<string, SkillTranslation> = {
321321
zhTitle: 'iFlow 心流搜索',
322322
zhDescription: '通过 iFlow Search API 提供网页搜索、图片搜索和网页抓取能力的社区 skill。',
323323
},
324+
'videodb': {
325+
// No zhName - keep the VideoDB brand name
326+
zhTitle: 'VideoDB - 面向视频 Agent 的感知、记忆与行动层',
327+
zhDescription: 'AI Agent 的感知、记忆与行动层。通过自然语言摄取、理解、搜索、编辑、转录、添加字幕、捕获、监控和流式播放视频与音频。',
328+
},
324329

325330
// === Security Skills ===
326331
'building-secure-contracts': {

0 commit comments

Comments
 (0)