ludovicianul

I’ve been working on a side project called Dochia for a couple of years now (initially released under different naming, but having a fresh start now). It’s an open-source CLI tool that automates negative and boundary testing for APIs.

Background for why I built it is simple. In microservices-heavy systems, API testing is repetitive. Every service exposes endpoints, and a big chunk of the testing is the same: invalid payloads, boundary values, how does it handle null, spaces, etc. I found myself redoing these checks over and over, which felt like wasted time. Dochia takes care of this “shared layer” of testing automatically, freeing up time for the scenarios that are unique to each service.

What it does:

- Reads your OpenAPI spec - Generates hundreds of smart, edge-case payloads - Runs them against your API - Outputs an HTML report with the results

So far I've been using it for every service I've built and have it as a quality gate to deal with the negative and boundary scenarios from the beginning. I've seen teams going through pretty embarrassing situations where they spend months building something cool, only to get bugs like: user added a trailing space to their name and the flow if failing.

Repo is here: https://github.com/dochia-dev/dochia-cli

And the website: https://dochia.dev

I’d really appreciate feedback, bug reports, ideas about why this could or could not work in your setup.

Just curious about how people are using AI agents for their personal (or commercial) use. And which models you are using.