Shelby FAQs

Question 1

What is Shelby MCP?

Accepted Answer

Shelby MCP is a local-first Model-Context Protocol (MCP) server designed to give AI agents secure, typed access to Shelby storage workflows. It's built with a strong emphasis on filesystem security and developer tooling.

Question 2

How does Shelby MCP ensure data security?

Accepted Answer

It features a strict `SHELBY_WORKDIR` sandbox that actively narrows safe paths, rejects path traversal and symlink escapes, and prevents uploads/downloads from reaching arbitrary machine paths, minimizing accidental data exposure.

Question 3

Does Shelby MCP support both development and real-world Shelby operations?

Accepted Answer

Yes, it offers a fully usable mock provider with local on-disk blob storage for robust development and testing, alongside a real Shelby provider wired to `@shelby-protocol/sdk` for genuine operations, though the latter is currently experimental.

Question 4

What file operations can AI agents perform using Shelby MCP?

Accepted Answer

AI agents can inspect provider and account state, manage sandbox scope, upload files (text, JSON, blobs) with stream awareness, list blobs, fetch metadata, download blobs safely, and verify blob integrity against local files.

Question 5

Is Shelby MCP compatible with different operating systems?

Accepted Answer

Yes, its command-line setup and execution flows are designed to be cross-platform, working reliably in environments like PowerShell, `cmd.exe`, Bash, Zsh, and most general Node.js setups.

Shelby

Shelby

주요 기능

사용 사례

주요 기능

사용 사례