Vaulken connects your AI tools to S3 storage via the Model Context Protocol. Read, write, search, and organize files with scoped permissions and full audit control.
Complete file management for your AI, from simple reads to semantic search and PDF extraction.
Read, write, append, patch, move, and delete files. Your AI can manage an entire knowledge base autonomously.
Exact substring search across all files, plus meaning-based vector search powered by embeddings.
OAuth 2.0 authorization with read-only or read-write scopes per client. Path-level access control included.
Connect any S3-compatible storage: Scaleway, AWS, MinIO, Cloudflare R2. Your files stay in your infrastructure.
Outline extraction, section-level editing, and frontmatter management. First-class support for Obsidian vaults.
PDFs are automatically converted to text via OCR. Your AI reads the content, not the binary. Results are cached for instant access.
Three steps to connect your AI to your files.
Bring your own S3 bucket or let Vaulken provision managed storage. Set permissions and path scoping from the dashboard.
Each vault comes with read-write and read-only MCP URLs. Copy the one you need from the dashboard.
Point Claude Desktop, Cursor, or any MCP client to your Vaulken server. Your AI can now read, write, and search your files.
Your files never touch our servers. Only irreversible vector indexes are stored for semantic search.
File contents stay in your S3 bucket. We never read, copy, or cache your files.
S3 keys are encrypted at rest with a hardware key management service. Never stored in plaintext.
All infrastructure runs in Paris, France. Your data never leaves the European Union.
Your data is never used to train or improve any model. No exceptions.
Free to start. One paid plan for everything else.
Perfect for getting started
For power users and teams