What it does
The Aito MCP server gives MCP-compatible clients access to Aito's prediction engine. It returns predictions over your own data together with a probability ($p) and an explanation ($why), and it does not require training a model. The publisher states when Aito does not fit a use case.
It is suited to adding a predictive feature to an application or agent, for example predicting a GL code for an invoice from its description. The packaging includes a Claude Code plugin with a skill, add-predictive-feature, which guides an agent through a fit check, building a schema from an existing table, a first prediction, choosing a threshold with _evaluate, and wiring act, assist and abstain outcomes.
The server is packaged alongside the Aito Python SDK, which is installed with pip and also provides a command-line interface. Running it requires an Aito instance. The documentation's examples use a public read-only sandbox, while production use needs your own instance URL and API key.
Limitations and requirements
Requires an Aito instance; production use needs your own instance URL and API key. The public sandbox is read-only. The publisher notes that Aito does not fit every use case.
Safety
Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.
- Runs code or changes an assistant's behaviour
Sources
- Official MCP Registry first seen 13 Nov 2026 · last seen 13 Nov 2026 · its page on the source
Aito
Predictions over your own data with $p and $why, no model training; says when Aito does not fit.
v1.3.0As shown on registry.modelcontextprotocol.io when we last read it