Build to Break: Your Guide to Cybersecurity at Hacktoberfest HackDay x OWASP JIS Univerity

A guide for the Hacktoberfest Hack Day Kolkata x OWASP JIS University hackathon outlines how participants should build focused cybersecurity projects, from secret scanning and IAM privilege-escalation analysis to detection engines, and recommends giving open-weight models like Gemma a bounded role such as explaining detections or recommending remediation. It advises treating AI output as untrusted input that must be validated before any controlled action, and stresses demonstrating real attack scenarios rather than only happy paths.

Welcome to Hacktoberfest Hack Day Kolkata x OWASP JIS University https://events.mlh.com/events/14953-hacktoberfest-hack-day-kolkata-x-owasp-jis-university 💖 Build a project that solves a real cybersecurity problem. Your project can focus on application security, cloud security, AI security, authentication, supply-chain security, detection and response, DevSecOps, or another security challenge. You don't need to build a complete enterprise security platform. Pick one meaningful problem, build a focused solution, and demonstrate that it works. Here are some areas to explore: These are starting points, not restrictions. If you have a different cybersecurity idea, build it. A simple way to structure a cybersecurity project is: ┌──────────────┐ │ Problem │ └──────┬───────┘ ↓ ┌──────────────┐ │ Attack │ └──────┬───────┘ ↓ ┌──────────────┐ │Security │ │Control │ └──────┬───────┘ ↓ ┌──────────────┐ │ Demonstrate │ │ & Test │ └──────────────┘ Ask: What are you trying to protect, and what are you protecting it from? For example: Problem: Exposed API credentials can be used to access sensitive resources. Attacker: Someone who obtains a leaked credential. Impact: Unauthorized access to data or services. Before building the solution, understand how the attack actually happens. Think about: Attacker → Attack Vector → Vulnerability/Weakness → Impact Attacker │ ▼ Stolen API Key │ ▼ Unauthorized API Access │ ▼ Sensitive Data Exposure Your project should intervene somewhere in this attack path. Your project should do something about the security problem . Source Code │ ▼ Secret Scanner │ ▼ Credential Detected │ ▼ Alert / Block Commit Or: IAM Policy │ ▼ Permission Analysis │ ▼ Privilege Escalation Path │ ▼ Finding + Remediation Security Logs │ ▼ Detection Engine │ ▼ Suspicious Activity │ ▼ Alert / Investigation The goal isn't to have the most features. A focused security control that works is more valuable than a large collection of unfinished features. If you're using Gemma or another open-source/open-weight model , give the model a meaningful role in your system. Security Detection │ ▼ Gemma │ ▼ Explanation / Correlation │ ▼ Remediation Recommendation Good uses for AI include: Treat model output as untrusted input , especially if it generates: Prefer: AI Output │ ▼ Validation │ ▼ Controlled Action rather than: AI Output │ ▼ Production Don't only demonstrate the happy path. Try to break your own system. At minimum, test: Normal Prompt ↓ Works Malicious Prompt ↓ Injection Detected ↓ Blocked Vulnerable Policy ↓ Privilege Escalation ↓ Detected ↓ Remediation Suggested Leaked Credential ↓ Detected ↓ Commit Blocked ↓ Developer Alerted A cybersecurity project should be demonstrated with an actual security scenario. A simple structure is: Attack → Detection/Prevention → Result Don't just show a dashboard and say: "Our system detects attacks." Show the attack. Show what your system detects or prevents. Show the result. Avoid making claims that you cannot demonstrate. Instead of: "Our system completely prevents phishing." Explain what your system actually does: "Our system detects suspicious websites using URL, DOM and visual features." "Our AI guarantees secure code." Say: "Our AI identifies potential security issues and suggests remediation." If you report metrics such as accuracy, explain how you measured them. OWASP Top 10:2025 https://top10.owasp.org/2025/ A starting point for understanding common application-security risks and finding project ideas. OWASP Top 10 for LLM Applications https://genai.owasp.org/llm-top-10/ Useful for projects involving LLMs, GenAI, AI agents, or open-source models. OWASP Cheat Sheet Series https://cheatsheetseries.owasp.org/ Practical implementation guidance for common security topics. OWASP API Security Top 10 https://owasp.org/API-Security/ Useful if your project involves APIs, authentication, authorization, or access control. OWASP Web Security Testing Guide https://owasp.org/www-project-web-security-testing-guide/ Use this to test your web application before submission. OWASP GSoC Project Ideas https://community.owasp.org/initiatives/gsoc/gsoc2026ideas Explore real OWASP open-source projects and technical problem areas for additional inspiration. OWASP Project Directory https://owasp.org/projects/ Browse existing OWASP projects across application security, AI security, cloud security, tooling, and more. These are provided for inspiration and further exploration . You do not need to contribute to an existing OWASP project for the Hackday. Before submitting, ask yourself: Problem → Attack → Security Control → Test → Result You don't need to build the next enterprise security platform. Pick one real security problem. Understand it. Build a focused solution. Try to break it. Then demonstrate that your solution works.