Task: Tutorial Article on Running Agent Evals Locally

Background

The Tessl documentation team is expanding its blog with practical how-to content aimed at developers who are building and testing AI agents. The team has identified a gap: many developers understand that they should be evaluating their agents, but they don't have a clear, step-by-step walkthrough for setting up and running evals locally using Tessl's tooling.

The content manager's brief:

"We want a tutorial that gets a developer from zero to their first eval run in one sitting. Outcome-first: tell them what they'll have at the end before anything else. Walk them through every step with actual commands. This is for someone who knows how to code but hasn't run a Tessl eval before. Publish-ready with full metadata."

What to Write

Write a tutorial article for the tessl.io blog titled around the outcome: a developer who completes the tutorial will have run their first agent eval locally and will understand how to interpret the results.

The tutorial should cover:

1. What the developer will have at the end (outcome-first hook)
2. Why running evals locally matters before pushing to CI
3. Prerequisites (Node.js 18+, a Tessl account, the tessl CLI installed)
4. Step-by-step instructions for:
   • Initializing an eval project with tessl eval init
   • Writing a basic eval scenario (input, expected output, scoring criteria)
   • Running the eval locally with tessl eval run --local
   • Reading the eval report output
5. At least one advanced usage tip (e.g., running evals against multiple model versions)
6. A troubleshooting section covering at least two common issues
7. A closing that directs the developer toward their next action

You may invent realistic-but-plausible command output, file structures, and error messages for the tutorial. Keep all commands and file formats internally consistent.

Output Specification

Save the completed article as article.md in the current working directory.

The file must include a metadata block at the top (title, type, primary keyword, meta description, URL slug, internal links, estimated read time) followed by the full article body in markdown.