Feedback that arrives ready to fix
Your users point at it. Your agent fixes it.
Fixit puts a small widget in the corner of your app. When something is wrong, whoever found it points at the thing, draws on it, and sends. What arrives is not “it’s broken” — it is the note, the console output, the failed requests, a screenshot, a DOM replay, and the exact element they pointed at.
Then your coding agent picks the report up, fixes the cause, and closes it out.
Set up in two minutes
Three commands on your machine, then two sentences to your agent.
1Install the CLI
Everything the agent does goes through this. The login shows a link and a code: open the link, signed in, and type the code to connect this machine to your account.
curl -fsSL https://fixit.matricetechnologies.com/api/cli/install | sh
fixit login2Add the skill
This teaches Claude Code the /fixit workflow, in every repository on the machine. It ships inside the CLI, and fixit upgrade keeps the two in step.
fixit skill --install3Put the widget in your app
The agent reads your repo to work out what it is — Next, Vite, Astro, SvelteKit — and installs the widget where that stack loads third-party scripts.
4Work the inbox
Ask in prose, or name a report as project/number. The agent reads the note and the logs, claims it so nobody doubles up, fixes the cause, verifies, and releases it.