MCP connector
What the Spera connector is
Connect Spera to Claude, Cursor, or VS Code and build strategies by asking. It works in your account, and it can never trade or touch your exchange keys.
Updated 2026-08-26·Rev. 2026.06
Most AI assistants can talk about trading strategies. They cannot open yours, change a parameter, run it over two years of candles, and tell you what happened.
The Spera connector closes that gap. Once connected, your assistant works inside your Spera account: it can find your strategies, edit them, compile them to check they are valid, backtest an exact revision, and write up what it found. You describe what you want in a sentence; the work lands in your workspace where you can open it and look.
Connect it from spera.bot/mcp — pick your app, approve once, restart, and ask.
What "MCP" means
MCP — the Model Context Protocol — is a standard way for an AI app to talk to an outside service. Think of it as a plug: Spera exposes a socket, and Claude, Cursor, or VS Code plugs into it. You approve the connection once, and after that your assistant has a set of Spera tools it can use on your behalf.
Nothing about it is Spera-specific. If you have connected any other MCP server to your AI app, this works the same way.
What it can do
- Find and read your strategies, modules, and custom nodes
- Create and edit them, then compile to confirm the graph is actually valid
- Backtest a specific revision and read back trades, equity, and metrics
- Write project knowledge and artifact documentation, cited to real results
It also carries eight guided workflows — build a strategy, run a backtest, document an artifact, and so on. These are not marketing labels: each one is a procedure your assistant follows, including which revision to pin, what evidence to attach, and when to stop and ask you instead of guessing.
What it cannot do
The connector never trades
There is no tool for deploying a bot, starting live trading, reading your exchange API keys, or moving money. Those operations are not merely restricted — they are not part of the connector's surface at all. An assistant cannot reach them by asking differently.
It also cannot reach anyone else's account. Everything happens under your own sign-in, and the permissions you approve apply only to your data.
Read-only until you say otherwise
A plain connection gets read access only — your assistant can look at your strategies but cannot change them. Writing requires you to approve a larger set of permissions, which you do on Spera's own sign-in screen. What you're approving walks through the list.
Next
- Connect your assistant — pick Cursor, Claude, VS Code, Claude Code, or Codex
- What to ask for — prompts that actually work
- When it doesn't connect — every failure mode and its fix