Tuning Engines

Lets an assistant fine-tune and manage custom LLMs on your data, run evaluations, and inspect governed runtime traces and approvals.

What you can do

A local CLI (tuningengines-cli) that exposes a built-in MCP server with 60+ tools. Through it an assistant can submit and monitor fine-tuning jobs on a GitHub repo, estimate costs, list and delete trained models, create datasets from S3, run evaluations, check inference usage, and inspect traces, policy decisions, approvals, and tenant registry metadata. It also offers governed skill invocation and agent messaging.

When to use it

Use it when you want an assistant to drive domain-specific model training and evaluation on your own data without managing GPU infrastructure, or to review governance, traces, and approvals for Tuning Engines workloads. It is desktop-only and aimed at developers already using the Tuning Engines platform.

Requirements

Runs locally via npm (npx or global install of tuningengines-cli); Node.js required. Needs a Tuning Engines account and the TE_API_KEY environment variable, obtained through the web app or 'te auth login'. Training, evaluation, and inference consume paid credits. Network access to the Tuning Engines API is required.