The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
AI Security Testing Platform (Sentinel Env)
About this work
As Large Language Models become increasingly integrated into production applications, ensuring their reliability and resilience against adversarial attacks has become a critical engineering challenge. Sentinel Env was developed as a modular AI evaluation platform that enables systematic testing of LLM-powered agents under realistic attack scenarios before deployment.
The platform simulates prompt injection, jailbreak attempts, social engineering, and other adversarial interactions through reproducible evaluation pipelines. Instead of producing only a binary pass/fail result, it measures robustness using structured scoring, resilience profiling, and detailed performance metrics that help identify security weaknesses and improve model behavior over time.
Designed with extensibility in mind, Sentinel Env exposes standardized API endpoints for evaluation workflows, making it suitable for continuous testing, benchmarking, and integration into AI development pipelines. The project demonstrates practical AI safety engineering by combining automated evaluation, reproducible experimentation, and quantitative model assessment into a unified framework.
Key Highlights
• Simulates prompt injection and adversarial attack scenarios
• Measures model robustness through resilience scoring and evaluation metrics
• Exposes reproducible API endpoints for automated AI testing
• Modular architecture designed for extensibility and benchmarking
• Enables continuous AI safety evaluation before production deployment
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Trending
Claude
Claude has entered the design space. How are you using it?
Contra University
Learn from expert creatives how to earn more using next-gen AI tools.
Brand Design
The best brand designers are on Contra. Scroll to see what's trending in brand design. What are you building?
creativeaiflow
Creative AI workflows are evolving. What tools do you use, and what are their strengths and weaknesses?
freelancerlife
Freelancer life is wins, pivots, and everything in between. What’s yours right now?
Related posts
Hi everyone! I'm Yuri, and this is my project for the Flowstep Challenge by Contra!
I have many volunteer friends around the world working in places where disasters have struck, and people need help. I often hear from them that they use different tools in their work, but they still really lack full automation to connect coordination centers with individual volunteers. This request inspired me to do this project!
I took a tool one of these volunteer organizations already uses and built out the parts it's missing:
- For coordinators (desktop) - all incoming requests with their statuses, warehouse state, the stats they actually need, and a map view of where every request is coming from.
- For volunteers (mobile) - a new request with its status, the full details, and a simple accept or decline.
Now, the fun part: how it was made. I ran Claude Code + Flowstep together over MCP. First, I wrote out a spec of everything the product needed to do and handed it to the setup. Then I went through the existing tool and documented its design system - this turned out to be the smartest thing I did, because it kept every screen consistent instead of feeling like ten different products. After that, I generated prompts in Flowstep, one dedicated prompt per user flow. I refined each screen with prompts, and honestly, most of the time it landed on the first or second try. The few tiny things left over I just nudged by hand right in Flowstep.
And I'm genuinely happy with how it came out. Doing all of this manually would've cost me a solid 6–8 hours. This took me a bit over two.
And the agent messages provided for the challenge were enough for me to finish. I was only 1 short - to adjust an icon on the last screen (hope you won’t blame for that).
🔗 The result of my work in Flowstep: https://app.flowstep.ai/file?activeFileId=0c294720-d873-4ced-b99c-e6f20f818209
Thank you all so much for checking it out, and I'd be grateful for your feedback and comments!
#FlowstepChallenge # #Flowstep #Claude #ClaudeCode #Weavy #UX/UI #ProductDesign #Saas #Dashboard
This CrisisCoord concept is brilliant.
The dashboard + mobile + warehouse flow feels so real and needed.
Love how you handled the urgency with the timers and priority system.
Great execution!
On a scale of 1–10, how much room is there to improve this using Envato?
3 voted
30%
7 voted
70%
10 votes
Closed
Depends on your undderstanding of aesthetic, but they look great btw!
"The first AI results didn't match my vision. They looked generic and lacked emotion. Instead of accepting them, I refined my prompt and asked for a tiger with a fiery appearance, a dark aura, and an intense expression that would make viewers feel intimidated just by looking into its eyes. After several attempts, I selected the version that best captured the mood I had imagined. The AI generated the image, but the creative direction, prompt refinement, and final selection were entirely my decisions."
