WayCast
Enables cloud-based voiceover generation with realistic narration.
Offers cloud-based copywriting for generating brand scripts and storyboards.
Supplies city background images for the brand film.
Supplies city background images for the brand film.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@WayCastturn https://stripe.com into a brand film"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
WayCast
Turn any company's URL into a polished vertical brand‑film — script, voiceover, city backgrounds, final MP4 — without opening an editor.
Open‑source · self‑hosted · local‑first. No avatars, no stock‑footage clichés — clean institutional motion‑graphics. The AI agent driving it is the writer, so it runs with zero cloud keys if you want.
中文说明 → README.zh.md · MIT OR Apache‑2.0 · Node ≥ 22 · pre‑alpha

60‑second try
npx waycastai doctor # what your machine can do
npx waycastai make https://stripe.com --reuse-bg wjdigital # URL → brand package in ./brands/
npx waycastai render <slug> --quality standard # → ./output/*.mp4Everything lands in your folder. Prefer a UI? npx waycastai console → http://127.0.0.1:4178.
Related MCP server: ai-video-mcp
What it makes
A 1080×1920, 8‑shot institutional brand film — cover → who‑we‑are → what‑we‑do → why → how → who‑it‑powers → beliefs → CTA — data‑driven from a reusable block library, EN + ZH, real‑voice narration, per‑shot city backgrounds.

Why WayCast is different
Most "URL → video" tools are paid SaaS, or wrap a cloud API behind talking‑head avatars. WayCast is the opposite end:
WayCast | |
Agent‑native | Driven by Claude Code / any agent via MCP — the agent writes the storyboard itself, so good copy needs no LLM key |
Local‑first | Renders locally; bundled local TTS (Kokoro EN / CosyVoice ZH same‑voice clone, Apache‑2.0) — or BYO cloud key |
Zero‑dependency | Pure Node + ffmpeg. No framework lock‑in |
Style | Clean text‑motion brand films — no avatars, no random stock B‑roll |
Bilingual | English + 简体中文 first‑class (most OSS video tools are EN‑only) |
Yours | Self‑hosted, dual‑licensed, keys stay in your |
Three ways to drive it
CLI / npx —
npx waycastai make <url>→render.Any coding agent — clone, point Claude Code at it; it reads
CLAUDE.mdand authors + renders the film itself. Seedocs/agent-usage.md.MCP —
claude mcp add waycast -- npx waycastai mcp→ 7 tools (waycast_scrape / write_brand / render / …). Seedocs/mcp.md.
How it works
URL ─▶ scrape ─▶ brief (+brand color) ─▶ script (EN/ZH copy) ─▶ pick city images
▼
brand package (brand.json + storyboard.json + vo.json + bg)
▼
local TTS ─▶ retime to narration ─▶ render (hyperframes) ─▶ MP4Every stage has a keyless local/deterministic fallback and an optional BYO‑key upgrade. Anti‑fabrication is enforced — the copy only uses facts from the site.
Backends (all optional)
Capability | Local (offline) | Cloud (BYO‑key) |
Voiceover (required, pick one) | Kokoro (EN) / CosyVoice (ZH) | OpenAI / ElevenLabs / Azure |
Copywriting | deterministic draft or the driving agent | Anthropic / OpenAI |
City images |
| Pexels / Unsplash |
Keys go in .env (.env.example). npx waycastai doctor shows what's ready. Full setup: docs/install.md.
Status — pre‑alpha (honest)
Core pipeline (URL → standard MP4) is verified end‑to‑end via the local path. Cloud LLM/TTS/image adapters are implemented but not yet live‑tested; a one‑click Docker image for the local models is WIP; Windows is untested (use WSL). It's a self‑hosted dev tool, not a turnkey consumer app. Changelog: CHANGELOG.md.
License
Dual‑licensed MIT OR Apache‑2.0 — your choice. Third‑party components (local TTS models, fonts, CC‑BY sample images, GSAP via CDN) in NOTICE.
Contributing
Early days — issues and ideas welcome. See CONTRIBUTING.md. If it's useful, a ⭐ helps others find it.
This server cannot be deployed
Maintenance
Related MCP Connectors
Create branded vertical video ads from ecommerce product URLs or product photos.
Turn URLs, recordings, docs, assets, or ideas into launch films, demo recordings, and decks.
Build, run, schedule, and publish AI video pipelines to YouTube and TikTok from any MCP client.
AI video producer: directs your idea or script into a finished, fact-checked video. MCP-native.
Related MCP Servers
AlicenseNot gradedqualityBmaintenancePhoto-to-reel MCP for solo founders and SMBs. Upload 1–10 photos and get a captioned vertical reel for Instagram, TikTok, YouTube Shorts, or Facebook — with motion, library-matched music, and optional AI voiceover.MIT- AlicenseNot gradedqualityDmaintenanceTurns text scripts into complete videos with stock footage, AI narration, and subtitles via MCP tools for AI assistants.MIT
- AlicenseNot gradedqualityBmaintenanceEnables turning any video into scored, captioned, ready-to-post vertical clips directly from your MCP client, with capabilities to download finished MP4s, schedule posts, and discover trending content.15 npmMIT
- AlicenseBqualityBmaintenanceEnables AI agents to take a raw script all the way to a finished, downloadable short-drama .mp4, covering AI rewrite, character consistency, storyboards, frames, video shots, TTS voiceover, and final cut with quote-before-spend billing from any MCP client.1321,529 npmMIT