TuiML is a self‑contained machine‑learning library built specifically for AI agents. Traditional libraries such as Weka and scikit‑learn target human programmers; agents can only use them by recalling APIs and generating code, which hides functionality, pushes errors to runtime, and loses experimental state between turns.
TuiML describes every component with machine‑readable metadata and parameter schemas, allowing an agent to search, inspect, compose validated workflows, and register new components that become discoverable later. Each call is validated, seeded, and traced, and sessions are exported as runnable notebooks, ensuring reproducibility by construction.
The Model Context Protocol (MCP) serves as a single specification layer that drives agent‑framework adapters, a Python API, a CLI, and local model serving, while data and models never leave the machine. Benchmarks show TuiML remains competitively predictive with scikit‑learn and Weka. To a human user it looks like a conventional library; to an agent it is a readable, extensible, and autonomous ML platform. TuiML is open source, with documentation at https://tuiml.ai.
Review