Free, unlimited large-file storage built on top of Discord's own attachment limits. Files (2GB–30GB videos, in practice) are streamed, split into chunks under Discord's per-message attachment cap, and uploaded as messages across a private server — nothing is stored anywhere else. A manifest tracks where every chunk landed so the file can be rebuilt and integrity-checked (SHA-256) on download. Controlled from CLI, a web dashboard, and (planned) desktop/Android clients.
Live: discvault.onrender.com
- Chunking — a file is streamed (never fully loaded into memory) and split into pieces sized to fit under Discord's free-tier attachment limit
- Upload — chunks are pushed to Discord as message attachments, parallelized without tripping rate limits, and resumable if an upload fails mid-way
- Manifest — a per-file record of every chunk's location (server, channel, message) is what makes reassembly possible
- Integrity — SHA-256 hashing verifies the rebuilt file matches the original, byte for byte
- Sharing — an invite code carries your server and channel IDs; a friend brings their own bot and everyone on the same channels sees the same files, no bot-token sharing
See docs/TECHNICAL_OVERVIEW.md for the full design rationale and protocol details.
| Sign in | Vault library |
|---|---|
 |
 |
| Chunked upload | Server settings, sharing |
|---|---|
 |
 |
| Layer | Technology |
|---|---|
| Core | packages/core — chunker, SHA-256 hashing, manifest schema, rebuilder (shared logic, TypeScript) |
| Discord I/O | packages/discord-adapter — REST client for chunk upload/download |
| CLI | apps/cli — discvault upload / download / list; on npm as discvault (npx discvault init) |
| Web | apps/web — Next.js dashboard + API, Discord OAuth |
| Deployment | Render.com |
packages/
core/ chunker, hashing, manifest, rebuilder
discord-adapter/ Discord REST client
apps/
cli/ discvault upload / download / list
web/ Next.js dashboard + API backend
docs/ technical writeups
render.yaml Render deployment config