Skip to main content

Guided project

Improve a coding agent through its harness

Take a small coding agent, build a fixed set of 20 tasks with automatic checks, measure a baseline, then change only the harness, one part at a time, and report what each change did to pass rate, cost and time.

Advancedabout 16 hours8 steps

On Agentic AI Engineer: Reliability

You will need

  • Python 3, Git, a GitHub repository and an API key with a spending limit you set yourself
  • A small open-source coding agent you can read in an afternoon, or a minimal agent loop of your own of about 200 lines
  • A budget in rupees for model calls, written down before you start

Lessons it uses

What you will show

  • A repository with the agent, the 20-task set and one command to run it
  • A results table with the baseline and each harness change, including cost and time
  • A CI job that runs part of the task set on every harness change
  • A one-page report on what helped, what did not and why

Sign up free to see every step

The 8 steps, the rubric to check yourself against, and ideas to go further. It is free, takes a few seconds and needs no card.

Accounts are for learners aged 18 and over. Under 18? The quick answer and the first part of every lesson stay open: browse lessons by topic.

Today's AI turns are used

Lessons, roadmaps, videos, practice cards and the code playground stay free with your account, with no limit.