Audio & videoCAN I VIBECODE
Shade?
BUILD FIRST
- One specific customer type and one primary use case
- A vertical slice that completes the main job without manual workarounds
- Visible limits, review steps, and a fallback when an external dependency fails
LEAVE FOR LATER
- Broad third-party integrations
- High-scale collaboration, syncing, and permissions
- Automation that can cause harm without human review
RESEARCH NOTES
What this tool actually does.
Shade (shade.inc) is a cloud NAS platform for media teams, marketed as an all-in-one replacement for LucidLink, Frame.io, Iconik, Dropbox, and Google Drive. It mounts cloud storage as a local drive so editors can stream and edit full-res files (MXF, R3D, BRAW, MP4) directly in an NLE without downloading or syncing. Features include AI-powered semantic search (natural-language queries, facial recognition, transcription), automated metadata extraction for 100+ file formats, automatic transcoding and preview generation, a review/approval suite with commenting and granular permissions (client collections, internal RBAC), and collaborative spreadsheet-style metadata views. It targets distributed creative teams from raw footage to final delivery. Note: a separate, unrelated product also named Shade (a local AI file browser in closed beta) exists; this summary covers shade.inc.
RESEARCHED BUILD PROMPT
Give this to your coding agent.
Build a focused replacement for Shade (shade.inc), the intelligent cloud NAS for media teams. Core workflow: a team uploads raw footage (MXF, R3D, BRAW, MP4) to a shared cloud workspace; the system mounts it like a local drive so editors stream full-res files into their NLE without downloading; it auto-extracts metadata, transcodes proxies, generates previews, runs AI tagging (faces, transcripts, semantic search), and supports client review/approval with comments and granular permissions.
User-facing requirements: (1) a web dashboard listing assets with auto-generated metadata and a spreadsheet-style view; (2) semantic search by natural-language query, face, and transcript; (3) review/approval with comments, shareable links, and role-based permissions; (4) a desktop client that mounts the workspace as a virtual drive and streams files on demand; (5) proxy/preview generation for common codecs.
Technical requirements: Node/TypeScript API; S3-compatible object storage for originals; Postgres + pgvector for tags, embeddings, and transcripts; an FFmpeg-based transcoding worker for proxies and previews; a WebDAV or FUSE virtual filesystem for the desktop mount; OAuth login with team/workspace scoping.
Implementation approach: monorepo with a React dashboard, Fastify/Express API, a BullMQ worker queue for transcode/transcription jobs, and an Electron or Tauri desktop client using a WebDAV/FUSE mount. Use off-the-shelf models (e.g., Whisper for transcription, a face-clustering model) behind an abstraction so models are swappable.
Do not copy Shade's brand, logo, marketing copy, proprietary assets, or UI design; build an original interface. Out of scope: real-time multi-user NLE collaboration, mobile apps, billing/payments, and production-grade distributed storage — stub these with clear interfaces. Deliver a README covering setup (env vars, Docker Compose for Postgres/Redis/MinIO), how to run the API, worker, and desktop client, and important limitations (mount performance, model accuracy, single-region storage).