Skip to main content
These guides cover the day-to-day mechanics of working with Shanone once you’re connected: how to phrase requests so your agent uses the meta-tools well, how to keep permissions sane as your team grows, and how to unblock yourself when something doesn’t work.

Best Practices

Get precise, efficient tool calls out of shanone_search_tools and shanone_execute_tool

Troubleshooting

Fix connection, authentication, and permission errors

MCP Overview

The meta-tool architecture in depth

Permissions & Access Control

How Root and SA2 users manage who can run what

​
What these guides assume

You’ve already got Shanone connected to at least one MCP client (Cursor, Claude Code, etc.) — if not, start with the Quickstart. Everything below is about how to use Shanone effectively once it’s wired up, not how to wire it up in the first place.

​
The mental model

Shanone doesn’t hand your agent a giant static list of tools. It hands it a handful of meta-tools that search a live catalog of 269+ integrations and execute whichever one turns out to be relevant: Most of the guidance in Best Practices is really about giving your agent enough signal to pick the right tool on the first shanone_search_tools call, instead of guessing and retrying.

​
Need help fast?

Documentation

Start from the top

Support

Contact the Shanone team