How to point an MCP client (Claude Code, Claude Desktop, or the Claude mobile app) at Cheatsheet, so it can search and manage your cheats and Tasks directly. If you're instead writing your own custom integration against the REST endpoints, see the API Reference.
On this page:
There are two ways to connect an MCP client to Cheatsheet. Both reach the same server and the same 30 tools, they only differ in where your API key ends up and which Claude surfaces they reach:
.mcp.json. Needs a one-time local setup and doesn't reach the mobile app, but keeps your key on your own disk instead of in another company's settings UI.Not sure which one? Start with Option A, it's the fastest to set up and needs no local files.
In claude.ai, Claude Desktop, or the Claude mobile app, go to Settings → Connectors → Add custom connector and enter this URL:
https://cheats.aarontrotter.com/mcp
Then choose how to authenticate:
Authorization: Bearer csk_live_...Since this key is typed into another company's settings UI rather than kept on your own disk, it's worth creating one scoped narrowly to this connector (read-only, if you don't need write access) so it can be revoked on its own.
Set up once, this is account-wide, available everywhere you're signed in to Claude.
For Claude Code, or any other client that reads a .mcp.json-style config, a small local process handles the connection so your API key stays in a file on your own disk instead of a client config or shell environment variable. Pick one:
/plugin marketplace add AaronTrotter/cheatsheet-mcp /plugin install cheatsheet@cheatsheet-mcpThen create a key from the User page's API Access section, save it to the file the plugin's error message points you to, and restart Claude Code to approve the
cheatsheet server when prompted. See the cheatsheet-mcp repo for details..mcp.json setup. A small script that reads your key from a gitignored .env.local file and forwards every call to the MCP Connector below. See Setting up the MCP proxy for the full script and step-by-step setup.local-mcp-server/ folder, useful as a starting point if you want to customize the tool set yourself. Covers the full tool set, see its own README.All three keep the key on your own machine and, like every local-process option, don't reach the mobile app, which has no local process to run one in.
Both options above ultimately talk to the same remote MCP server:
URL: https://cheats.aarontrotter.com/mcp Header: Authorization: Bearer csk_live_...
Exposes 30 tools over the same key/quota/tier rules as the REST API. Read-scoped: search_cheats, get_cheat, get_revisions, search_tasks, get_task, search_brain, get_brain, search_pennies, get_penny_summary, search_projects, search_project_tasks, get_project_task, get_guides. Write-scoped: add_cheat, update_cheat, delete_cheat, add_task, update_task, delete_task, add_brain, update_brain, delete_brain, add_penny, fill_penny, void_penny, add_project, add_project_task, update_project_task, move_project_task, delete_project_task. All but get_guides map directly to one of the REST endpoints documented on the API Reference page, which also covers the optional X-Cheatsheet-Client header for per-client usage tracking, it applies here too.
The task and brain tools cover two separate sections, not one list with a filter. The task tools reach your Tasks page (note and list); the brain tools reach your Brain page (brief, rules and memory). Neither can see or change the other's items, so an assistant tidying up your notes can never touch the rules you wrote for it, and vice versa.
The tool list is narrowed to what your key can actually use, which is worth knowing because a shorter list makes a client better at picking the right tool. A read-only key is offered only the 13 read tools: the write tools could never do anything but tell you the key lacks write access. A key limited to particular sections (ticked when you create it on the User page) is offered only those sections' tools, so a read-only key for Cheats alone is offered 3 tools rather than 30. If you just want Cheatsheet available as reference material, that is the tidier connection as well as the safer one.
get_guides returns the brief and rules entries from your Brain. You don't normally need to ask for it: the same guidance is sent to the client automatically as it connects, so an assistant has read your rules before it does anything. The tool is there for re-reading them later in a long session, or after you have changed them.
You won't normally need to wire this up by hand — the Claude Code plugin and the MCP proxy above both do it for you. If you're configuring a client's .mcp.json yourself instead, here are the two raw entry shapes. Through the proxy, so the key stays in a file (see Setting up the MCP proxy for the setup this depends on):
{
"mcpServers": {
"cheatsheet": {
"command": "node",
"args": ["live-proxy.js"],
"env": {
"CHEATSHEET_ENV_FILE": ".env.local"
}
}
}
}
Or direct, with no local process, if you're fine putting the key in a real shell environment variable instead:
{
"mcpServers": {
"cheatsheet": {
"type": "http",
"url": "https://cheats.aarontrotter.com/mcp",
"headers": {
"Authorization": "Bearer ${CHEATSHEET_API_KEY}"
}
}
}
}
For a "type": "http" entry, Claude Code can only expand ${...} from a real environment variable already set before it starts, it has no way to read one out of a file for this entry type. Either way, create the key first from the User page's API Access section, choosing read-only or read + write depending on what you want the client to be able to do, and restart Claude Code (or start a new session) after adding the config to approve the cheatsheet server when prompted.