WonderwallAi
"Your LLM's last line of defence against the internet."
3 followers
"Your LLM's last line of defence against the internet."
3 followers
WonderwallAi is an open-source AI firewall SDK that protects your LLM apps from prompt injection, jailbreaks, data leaks, and off-topic abuse. Three lines of code, under 2ms latency, zero external API calls. Catches API keys and personal data leaking in responses, strips metadata from uploaded files, and keeps your chatbot on-topic so users can't break it. MIT licensed and self-hostable. Hosted version available with dashboard, audit logs, and team controls. pip install wonderwallai.
Hi Product Hunt!
I'm Louis, solo founder of Skint Labs. Today I'm launching WonderwallAi, an open-source AI firewall SDK I built because LLM apps are way too easy to abuse.
The problem is simple. Once you ship a product with an LLM in it, bad actors find ways in fast. Prompt injection, data leaks, jailbreaks, off-topic abuse. Most teams find out the hard way after something embarrassing ends up in a screenshot on X.
WonderwallAi sits between your users and your LLM and blocks four things. It catches prompt injection before it reaches the model, enforces topic boundaries so people can't jailbreak it, scans responses for API keys and personal data leaking out, and strips EXIF and GPS metadata from any uploaded files. It runs locally, makes zero external API calls, and adds under 2ms of latency.
It's MIT licensed. Run pip install wonderwallai and you're protected in 3 lines of code. There's also a hosted version with a dashboard, audit logs, and team controls if you'd rather not run it yourself.
Product Hunt exclusive: use code PRODUCTHUNT at checkout for an extra 75% off! Valid until Sunday June 3rd.
50% EARLY_BIRD_50 offer runs for the first 300 customers .
Would love your feedback, especially from anyone who's been burned by prompt injection in production. Ask me anything, happy to dig into the technical side. ✌️