npm.io
0.3.0 • Published 3d agoCLI

@talocode/videolane

Licence
MIT
Version
0.3.0
Deps
0
Size
331 kB
Vulns
0
Weekly
0
SummaryDependencyVersions

VideoLane

Turn product workflows into publish-ready videos.

Open-source agentic video production engine for product demos, tutorials, and launch videos — turn scripts and product workflows into recorded, edited, captioned, and publish-ready video assets.

Why VideoLane

Agents should not only write scripts. They should help record, edit, package, and publish product videos.

VideoLane is a local-first tool that takes a script or production package and generates:

  • A structured video plan with scenes
  • Browser recording via Playwright (or dry-run/manual mode)
  • Captions (SRT, VTT, JSON)
  • Background audio
  • Shorts cutdowns
  • Thumbnail brief with text options
  • YouTube metadata (title, description, tags, pinned comment)
  • A packaged YouTube upload folder

How It Works

Script / production package
        ↓
VideoLane plan
        ↓
Browser recording / screen capture
        ↓
Captions + audio + render
        ↓
Shorts + thumbnail brief + metadata
        ↓
YouTube package

Install

npm install -g @talocode/videolane

Or run without installing:

npx @talocode/videolane@latest demo

Python:

pip install talocode-videolane

Quickstart

# Initialize a project
videolane init --template tera-tutorial --name my-video

# Generate plan from a script
videolane plan ./my-video/script.md --out ./my-video/plan.json

# Dry-run browser recording
videolane record --plan ./my-video/plan.json --dry-run

# Generate captions
videolane captions --script ./my-video/script.md --out ./my-video/captions.srt

# Generate background audio
videolane audio --duration 30 --style cinematic --out ./my-video/music.wav

# Generate YouTube metadata
videolane metadata --plan ./my-video/plan.json --title "My Video" --cta-url "https://example.com" --out-dir ./my-video/metadata

# Package for YouTube
videolane package-youtube --video ./my-video/final.mp4 --metadata-dir ./my-video/metadata --out ./my-video/youtube-package

# Run the full demo
videolane demo

Script-to-Recording Pipeline

VideoLane parses markdown scripts and production packages into structured video plans:

  • Detects scenes/sections from headings
  • Extracts prompts from code blocks
  • Estimates duration from word count
  • Generates browser action sequences
  • Creates YouTube metadata drafts
videolane plan docs/video1-production-package.md --out plan.json

Browser Recording with Playwright

videolane record --plan plan.json --url https://example.com --headless

Requires Playwright:

npm install playwright
npx playwright install chromium

If Playwright is not available, VideoLane falls back to dry-run mode showing the actions that would be executed.

ffmpeg Rendering

videolane render --recording raw.mp4 --captions captions.srt --audio music.wav --out final.mp4

Features:

  • H.264 + AAC encoding
  • Caption burn-in
  • Audio mixing
  • Multiple aspect ratios (16:9, 9:16, 1:1)

Captions

videolane captions --script script.md --out captions.srt --format srt

Modes:

  • script: Split narration into timed captions (no API key needed)
  • timed: Use scene timestamps from plan
  • transcribe: Use OpenAI/Tera API (requires API key)

Background Audio

videolane audio --duration 60 --style ambient --out music.wav

Styles: ambient, cinematic, minimal, study, tech

Generated locally — no API keys needed.

Voiceover (TTS)

VideoLane generates voiceover narration from plan scenes using edge-tts.

# Generate voiceover from plan
videolane voiceover --plan plan.json --out voiceover.wav

# Render with auto voiceover (default)
videolane render --plan plan.json --out final.mp4

# Disable auto voiceover
videolane render --plan plan.json --out final.mp4 --no-voiceover

Voices: en-US-GuyNeural (default), en-US-ChristopherNeural, en-GB-RyanNeural, en-AU-WilliamNeural

Requires: pip install edge-tts

Shorts Generation

videolane shorts --video final.mp4 --plan plan.json --out-dir shorts/

Generates a shorts plan and exports clips using ffmpeg.

Thumbnail Brief

videolane thumbnail --plan plan.json --title "My Video" --style dark --out thumbnail.json

Generates:

  • Thumbnail brief with text options, colors, layout
  • SVG mockup
  • Canva instructions

YouTube Metadata

videolane metadata --plan plan.json --title "My Video" --cta-url "https://example.com" --utm-campaign my_video --out-dir metadata/

Generates:

  • Title options
  • Description with UTM link
  • Tags
  • Pinned comment
  • Chapters
  • Upload checklist

YouTube Publishing Package

videolane package-youtube --video final.mp4 --metadata-dir metadata/ --out youtube-package/

Creates a folder with everything needed for manual YouTube upload.

YouTube Upload Safety

VideoLane can upload videos to YouTube, but upload actions are protected by explicit confirmation flags.

# Package only. No upload.
videolane youtube upload --video final.mp4

# Upload as unlisted.
videolane youtube upload --video final.mp4 --confirm-upload

# Public upload requires an extra confirmation.
videolane youtube upload --video final.mp4 --confirm-upload --privacy public --confirm-public

# Preview only.
videolane youtube upload --video final.mp4 --dry-run

Safety model:

Command What happens
videolane youtube upload --video v.mp4 Package only — no upload
videolane youtube upload --video v.mp4 --confirm-upload Upload as unlisted
videolane youtube upload --video v.mp4 --confirm-upload --privacy public Blocked — needs --confirm-public
videolane youtube upload --video v.mp4 --confirm-upload --privacy public --confirm-public Upload as public

Setup:

  1. Enable YouTube Data API in Google Cloud Console
  2. Create OAuth 2.0 credentials (Desktop app type)
  3. Run OAuth flow to get refresh token
  4. Save as youtube-credentials.json in project or ~/.videolane/

Credentials are auto-discovered from:

  • --credentials <path> flag
  • ~/.videolane/youtube-credentials.json
  • /workspace/.youtube-credentials.json
  • YOUTUBE_CREDENTIALS_PATH env var

Templates

videolane init --template product-demo --name demo
videolane init --template youtube-tutorial --name tutorial
videolane init --template launch-video --name launch
videolane init --template shorts-demo --name shorts
videolane init --template tera-tutorial --name tera

CLI Commands

Command Description
videolane init Initialize a new project
videolane plan Generate plan from script
videolane record Record browser actions
videolane capture Capture screen/browser
videolane render Render video with ffmpeg
videolane voiceover Generate TTS voiceover
videolane captions Generate captions
videolane audio Generate background audio
videolane shorts Generate shorts cutdowns
videolane thumbnail Generate thumbnail brief
videolane metadata Generate YouTube metadata
videolane package-youtube Package for YouTube
videolane youtube upload Upload to YouTube (safe, requires confirmation)
videolane demo Run deterministic demo
videolane doctor Check dependencies
videolane serve Start API server
videolane mcp Start MCP server
videolane config Manage configuration

SDK

import { VideoLaneClient } from "@talocode/videolane";

const client = new VideoLaneClient();

const { project, plan } = await client.initProject({ name: "demo" });
const result = await client.plan({ path: "./script.md" });
const captions = await client.generateCaptions({ script: "./script.md", out: "./captions.srt" });

API

videolane serve --port 3110

# Health check
curl http://localhost:3110/health

# Generate plan
curl -X POST http://localhost:3110/v1/videolane/plan \
  -H "Content-Type: application/json" \
  -d '{"input": "./script.md"}'

MCP Tools

VideoLane provides MCP tools for AI agents:

  • videolane_plan — Generate video plan
  • videolane_record_browser — Record browser actions
  • videolane_render — Render video
  • videolane_generate_captions — Generate captions
  • videolane_generate_audio — Generate background audio
  • videolane_generate_shorts — Generate shorts plan
  • videolane_generate_thumbnail — Generate thumbnail brief
  • videolane_generate_metadata — Generate YouTube metadata
  • videolane_package_youtube — Package for YouTube
  • videolane_youtube_upload — Upload to YouTube (safe, requires confirmation)
  • videolane_doctor — Check environment
  • videolane_demo — Run demo

Python Package

pip install talocode-videolane
videolane-py demo
videolane-py plan --input script.md
videolane-py metadata --plan plan.json --title "My Video"

Integrations

  • Playwright — Browser recording (optional)
  • ffmpeg — Video rendering (optional)
  • ScreenLane — Screen capture (optional)
  • SkillLane — Skill export (optional)
  • Tera — AI content generation (optional)

Storage

Local storage at ~/.videolane/:

~/.videolane/
  config.json
  projects/
    <project-id>/
      videolane.json
      plan.json
      recordings/
      renders/
      captions/
      thumbnails/
      metadata/
      shorts/

Security / Privacy

  • Local-first — no data leaves your machine by default
  • No secrets required for core functionality
  • No YouTube credentials stored unless explicitly configured
  • Browser profiles kept local
  • API keys only used when you provide them

Talocode Ecosystem

Product Purpose
ScreenLane Screen capture and context
SkillLane Skill workflow structure
VideoLane Video production (this package)
ClipLoop Product update promo content
Tera AI assistant for students/builders
Codra AI code editor
SearchLane Search API
VectorLane Vector search
XProLane X Pro setup assistant

Demo Video

Run the demo:

videolane demo

This generates a complete demo output with plan, captions, audio, metadata, and YouTube package.

Roadmap

  • Full Playwright browser recording
  • ffmpeg caption burn-in
  • Background music generation
  • Shorts auto-cut
  • YouTube Data API upload
  • ScreenLane integration
  • Multi-language captions
  • Transcript-based captions (Whisper)
  • Thumbnail image generation
  • Video templates library

Limitations

  • v0.1 packages publish-ready assets; it does not fully replace human review
  • YouTube auto-upload is not enabled by default
  • Browser recording depends on Playwright and browser availability
  • Rendering depends on ffmpeg
  • Automatic login is not handled; use manual-login or browser profile
  • Captions can be generated from scripts locally; transcription requires provider keys
  • Thumbnail generation is brief-first; final thumbnail design may need Canva/editor review
  • Background audio is generated locally and should be reviewed before publishing
  • Complex editing still needs human review

Contributing

Contributions welcome. Open an issue or PR at github.com/talocode/videolane.

License

MIT Talocode

Keywords

videoyoutubescreen-recordingplaywrightffmpegcaptionsai-agentsproduct-demotutorialsmcpdeveloper-toolstalocodeclisdk