BreakMyAgent – Open-source red-teaming sandbox for LLM system prompts
LLM-as-Judge red-teaming for system prompts, but Anthropic/OpenAI already ship this internally.
Drag components onto a canvas and generate AI-powered layout and style variations — right inside your Next.js project.
The repo is here: https://github.com/B1u3B01t/design-playground.
We've been working together for six years. Last year we built an AI design agent because we thought designers were the biggest bottleneck. But we were wrong.
After talking to teams, we realised designers, PMs, founders and even sales people all wanted to quickly change and explore flows for their users, that looked and behaved like the real app.
But devs don't trust non-technical teams to work on production and not fuck up the code. Agree. So the other way is setting up another mock repo.
And that defeats the point. It adds maintenance cost of keeping both in sync. This is what happened when both of us were working together too. So we built an open-source Playground.
It has all the features out of the box: super intuitive, good ux, links to claude/codex, latest features, easy additions of skills, collaborative and best of all, safe guardrails so if you want, you can get the code in one click (unlike figma or other closed source shit).
I work on the Playground. Rahul on Conductor. We're super claude pilled and believe in this duo being the new product team size, something coinbase and square are already doing with "pods".
Right now it supports Nextjs projects. We will definitely add more frameworks and improve the developer experience. We'd especially love feedback from teams already maintaining internal sandboxes.
Go let your non-tech team play.
LLM-as-Judge red-teaming for system prompts, but Anthropic/OpenAI already ship this internally.
Docker-sandboxed agent orchestration for red teams joins a crowded automated pentesting space.
Guardrailed sandbox sessions beat Loombloom with AI operator walkthrough modes.
Open-source alternative to Navattic for creating interactive product demos.
It bundles knowledge base articles, a public roadmap with voting, structured feedback boards, release notes and API reference into one branded portal — exactly the convenience product teams ask for when they're juggling GitHub issues, Canny and a separate docs site. The admin UI in the screenshot (analytics, content counts, role controls) reads clean and usable, but this lives in a crowded field; success will depend on smooth imports, strong integrations and demonstrable SEO/search benefits.
Curated AI video directory, but exposed database credentials in page source.