July 16, 2026
Bot fixes botched bots
Show HN: Libretto PR agents – Automatically fix failing playwright scripts
AI says it can fix your broken test scripts, and the crowd instantly asked: why not DIY
TLDR: Libretto launched a tool that watches for broken website test scripts and opens a code fix automatically. The biggest reaction wasn’t awe but skepticism, with commenters immediately asking whether this is truly a product or just a reusable prompt anyone could copy.
A new Show HN post is pitching a very modern fantasy: your website-checking script breaks, an AI helper investigates the page, and then opens a GitHub pull request with a fix for you. In plain English, it’s a bot mechanic for busted browser tests. Libretto says you keep your current setup, only call the agent when something fails, and it will inspect the live page, suggest a code change, and even package it neatly as a proposed update. It also waves a big crowd-pleaser flag: free to use, open source, and you can bring your own AI provider key.
But the real fireworks came from the comments, where the first big reaction was basically: hold on, what exactly am I paying attention to here? One early commenter, pryelluw, delivered the classic tech-world side-eye: if the prompt is visible, why can’t people just copy the idea and make their own version? That instantly turns the launch into a familiar internet drama: product magic or just a nicely wrapped prompt with extra steps?
That tension is the whole vibe. Supporters will see a time-saving assistant that patches annoying breakages automatically. Skeptics hear "AI auto-fix" and smell a glorified shortcut wearing startup clothes. The humor writes itself: developers already imagine a bot opening a pull request to fix another bot’s mistake while everyone in the comments argues over who really did the work. In other words, the tool promised fewer headaches, and the community responded by generating a fresh one for free.
Key Points
- •Libretto introduced a PR agent that investigates failed Playwright scripts and opens GitHub pull requests with proposed code fixes.
- •The tool is designed to be added at the failure boundary so existing Playwright fixtures, retries, logging, deployment, and workflow structure can remain unchanged.
- •It works with local, self-hosted, and hosted browser providers as long as a live Playwright Page remains available during debugging.
- •Users bring their own LLM provider API keys and browser infrastructure, and Libretto says it does not charge for the PR agent itself.
- •The current package supports Playwright only; Selenium and Puppeteer are not yet supported, and the Playwright debugger package is open source under the MIT license.