Vaibhav Bhandari
Kavach
Kavach wraps any AI agent in a sandbox, a policy gateway and a secrets vault from one config file, so a prompt-injected agent can't leak card numbers or API keys.
Pi
View project Cyberdefense Hackathon
Project gallery
Explore the projects and technologies built at the hackathon.
Showing 1–30 of 78 projects
Vishal Juneja
0DayForge -- Every push is investigated, PoC proven, and patched.
0DayForge is an autonomous security remediation harness.
Guild.aiClickHousePiSemgrep
View project 
Thomas Llamzon
StackWatch
StackWatch is an agent platform that develops your organization's entire emerging threat exposure given only the domain.
SemgrepClickHouseSenso.ai
View project C
vipin211
Coding Logs
Watches the session logs of AI coding agents (Claude Code, Codex) and shows what they did on a developer's machines: destructive commands, leaked secrets, permission bypasses and guessed decisions. Each finding becomes a rule or fix the developer approves.
Guild.aiClickHouseSemgrep
View project 
divya chandana
TrustMeBro
TrustMeBro is a Claude Code plugin that blocks vulnerable code before your AI agent writes it, and learns from every catch so it never lets the same mistake through again.
SemgrepClickHouse
View project 
Anonymous builder
Ghost Ops
Ghost Ops is an autonomous AI counterintelligence platform that detects rogue agents, monitors suspicious behavior and memory manipulation, and blocks unauthorized actions in real time using secure MCP gateways, automated threat investigations, and verifiable forensic evidence.
Guild.aiClickHouseAkash
View project 
bk
Voice of the People
Stop AI-cloned voices from defrauding your dad
Hugging FaceClaude CodeForensics 0.3BFish S2 Pro
View project 
securitydev
Aegis: Autonomous Defense Loop
Aegis reads a deploy you already ship, joins misconfigurations into attack paths, hardens the files, and rescans the workspace and the published URI to prove the path is closed.
Guild.aiAkashSemgrep
View project 
Noel Bobby
ProofLock
ProofLoop independently tests AI-generated security fixes and backs its verdicts with reproducible evidence.
ClickHouseAkashSemgrep
View project Gayathri Ramakrishnan
Antibody
An autonomous agent that finds a real SQL-injection bug with Semgrep, fixes it with Guild AI, verifies the fix via independent tests, then deploys the patched app live and logs tamper-evident proof to ClickHouse — closing a security hole without closing the business.
Guild.aiClickHouseAkashSemgrep
View project 
ashish ranjan
Memory Quarantine Gateway
Preventing Data leaks and Breaches from reaching a shared company brain.
ClickHouseAkashSenso.ai
View project Makarand Bhalerao
Attack-fix-verify
An autonomous security agent that attacks a running API to find real vulnerabilities, uses an LLM to patch them, and crucially proves each fix by re-running the exact exploit before opening a GitHub pull request
Semgrep
View project 
karan
PatchMedic
PatchMedic: an autonomous security engineer that can’t be turned into the attack
ClickHousePiAkashSemgrep
View project 
rotemb
Parry the Appsec AI SOC
Parry is an autonomous AI SOC for web apps. It watches every HTTP request, detects live attacks as SQL over ClickHouse, triages and patches them with an LLM on Akash GPUs, blocks the attacker, rotates exposed secrets, and writes its own detection rules — no human in the loop.
Guild.aiClickHouseAkashSemgrep
View project 
NarasingaMoorthy V
Immune
The agent security loop that turns every attack into an antibody
Guild.aiClickHouseSemgrepSenso.ai
View project 
skinemtowinem
gtmsos-personal
Agentic Development Lifecycle (ADL) – Autonomous DevSecOps Engine
ClickHouseSemgrep
View project 
Bindu Bhargava Reddy Chintam
Tripwire
The immune system for AI-agent fleets. It holds risky agent actions before they run, traces a poisoned input across the fleet, and cures it with a fix proven on ~30M events in ClickHouse.
Guild.aiClickHouseAkashSemgrep
View project Om Joshi
Agent Honeypot
It is a detection infrastructure that is supposed to reveal which AI agents have already been compromised by prompt injection, by logging every time one takes planted bait.
SemgrepClickHousePi
View project 
Anonymous builder
Hedgerow
Hedgerow is a plug-in security appliance that watches your home network, explains suspicious activity, and uses AI agents to make approved repairs and verify they worked.
ClickHouseSenso.aiSemgrep
View project 
Naman Kumar
threatwatch
ThreatWatch continuously scans a GitHub repository for newly published security threats, matches them to installed dependencies and code evidence, and provides verified remediation guidance.
ClickHouseSemgrepAkash
View project 
Derek Peng
Hive
I hid a prompt injection in a security blog telling AI to report a fake breach. My agent swarm read it, ClickHouse caught it, and the fake breach never shipped.
Senso.aiSemgrepAkashClickHouse
View project Nayeon Shin
Takedown Orchestrator
An autonomous cyberdefense agent that turns suspicious URLs into evidence-backed reports and verified outcomes, helping security teams move from detection to response.
ClickHouseAkashSemgrep
View project 
Rithik Sarvesh B
Athena
Evidence-driven vulnerability remediation - One solution for all the processes from threat intelligence, asset correlation, remediation and rescanning
ClickHouseAkashSemgrep
View project 
Paras Jain
Tripwire
Tripwire detects and contains risky AI agent actions in real time, then turns their audit trails into actionable security insights.
ClickHouseAkashSemgrepSenso.ai
View project 
Rishit Agrawal
DriftGuard
DriftGuard automatically detects API changes and security vulnerabilities through monitors on developer docs, finds affected code, and opens tested GitHub issues to alert developers.
Guild.aiClickHouseAkashSemgrep
View project 
Am L
Reva
Reva, security at the speed of your coding agent: a novel AI detection model, grounded in your project’s threat model and built on Semgrep, screens every function your coding agent saves, proves real vulnerabilities by running the code, and delivers a verified patch.
ClickHousePiAkashSemgrep
View project 
Nicholas Frangia
signal-integrity-live
Signal Integrity Live is an AI-assisted cybersecurity platform that detects schema drift in security telemetry, diagnoses broken detection rules using Google Gemini, validates proposed repairs through deterministic regression testing, and records audit events in ClickHouse to improve detection relia
ClickHouse
View project Rachna Pandya
Agent-assurance-lab
Continuous adversarial security testing for AI agents.
Guild.aiClickHouseSemgrep
View project 
Amir Khabaza
Aegis
Aegis gives on-call security teams a voice assistant that detects an active attack, proposes a fix, and deploys the patch only after a person approves it out loud
MongoDB AtlasOpenAI Agents SDKOpenAI Codex CLIClaude Sonnet 5 on OpenRouter
View project Hardik Goyal
Differential Swarm - Self Improving Security Evals
Differential Swarm is a set of dynamic security evals that runs on every PR: a swarm of red-team agents attacks the old and new builds in paired sandboxes and flags any regression in the agent’s safety.
OpenAI Codex CLISemgrepClickHousePi
View project