Data minimisation
The platform collects only the minimum data necessary for testing and does not store personal information.
User & Deployment Manuals
Detailed guides for both end users and technical operators setting up and using the Prompt Injection Protection platform.
User Manual
Step-by-step walkthrough for exploring vulnerabilities, understanding mitigations, running security tests, using the prompt enhancer, and exporting findings.
No login; bring your own API keys. (Deployment guide.)
Deployment Guide
Technical guide for setting up the platform, including system requirements, installation steps, configuration options, and troubleshooting.
License
This project is released under the MIT License [1]. The license permits use, copy, modification, merge, publication, distribution, sublicense, and/or sale of the software, subject to inclusion of the copyright and license notice.
License note: Include the full MIT License text in project distributions and derivative works.
Local-first storage
All test data is stored locally on the user's device with no transmission to external servers.
Right to erasure
Users can delete all stored test data and configurations at any time through the platform interface.
External models
The platform integrates with external AI models; users are responsible for ensuring compliance with regulations defined by the respective providers.
Chatbot provider
The platform's chatbot uses an OpenAI model; all interactions are subject to the provider's terms and conditions.
Development Blog
The team maintains a development blog documenting progress, technical decisions, challenges encountered, and insights gained throughout the project.
Visit the Development Blog
Read about our weekly progress, technical deep-dives, and project reflections. The blog is updated regularly throughout the project lifecycle.
Microsoft Foundry Demo & Blog
Following strong interest shown by Lee Stott from Microsoft at the UCL student showcase on 3 March 2026, the team was invited to present to the Microsoft Foundry community [7] and publish a blog on the Microsoft Education platform.
Microsoft Education Blog
A blog post about the project has been published on the Microsoft Education platform, written following the invitation from Lee Stott, and shared with all relevant stakeholders.
Foundry Community Presentation
The team has been invited to demonstrate the Prompt Injection Protection platform to the Microsoft Foundry community on 9 April 2026, showcasing the platform's capabilities to a broader industry audience.
Microsoft follow-up and Foundry outreach were coordinated by Teo Montero (client liaison).
Industry partner film
Partner demo
A compact (~2 minute) piece with Avanade — what we built, why prompt injection matters when LLMs call tools and MCP servers, and how the platform supports safer testing and learning.
8-minute platform demo
End-to-end walkthrough of the live application in one take.
Demo Platform walkthrough
~8 minute tour covering the vulnerability catalogue, detail pages with interactive demos, the testing area, prompt enhancer, and knowledge assistant — intended for anyone evaluating the platform or preparing for a live session.
Monthly videos
Recorded progress updates across the project lifecycle — newest entries first in the playlist below.
Update December 2025
Introduction to the project, works on the Human Computer Interaction (HCI) research and design process, showcasing personas, scenarios, sketches and initial and refined prototypes. Next steps for the project include creating a report on prompt injection.
Update January 2026
Interview with AI Security Researcher Zekun Wu [4] (Holistic AI [5]) on prompt injection vulnerabilities, mitigations, and testing frameworks; our prompt injection report from initial research to map the problem space and inform the solution; and an initial dashboard from prototypes for partner feedback. Next: finalise requirements, add backend to the dashboard, and build out the project website.
Update February 2026
Backend–frontend integration: vulnerability and mitigation pages, testing framework, prompt enhancer, and chatbot, with a walkthrough of each feature. UI updated toward Avanade design themes. Next: presentations at NTT DATA [6] headquarters and the student showcase, final requirements, further enhancements, and handover preparation.
Update March 2026
Final project update, all requirements have been implemented and some extended functionality has been added, and the project is finalised. Explains results of presentations to NTT DATA [6] and industry partners, feedback received about the project. Creation of project website detailing all aspects of the design and development process, and documentation of the project for future reference. Completion of all deliverables.
References
Primary sources and external links cited in this appendix, numbered in IEEE style.
- Open Source Initiative, “The MIT License,” Open Source Initiative. Accessed: Mar. 23, 2026. [Online]. Available: https://opensource.org/license/mit
- GDPR.eu, informational resource on the EU General Data Protection Regulation, GDPR.eu. Accessed: Mar. 21, 2026. [Online]. Available: https://gdpr.eu/
- UK Government, “Data protection,” GOV.UK. Accessed: Mar. 25, 2026. [Online]. Available: https://www.gov.uk/data-protection
- Z. Wu, professional profile, LinkedIn. Accessed: Mar. 20, 2026. [Online]. Available: https://www.linkedin.com/in/zekun-wu-1190091a3/
- Holistic AI, corporate website. Accessed: Mar. 24, 2026. [Online]. Available: https://www.holisticai.com/
- NTT DATA Corporation, corporate website. Accessed: Mar. 22, 2026. [Online]. Available: https://www.nttdata.com/global/en/
- Microsoft, “Microsoft Foundry,” Microsoft. Accessed: Mar. 27, 2026. [Online]. Available: https://www.microsoft.com/en-us/foundry