# OODA-Aegis

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Nikhil Kumar](https://cerebralvalley.ai/u/nobleimposter)
- **Demo video:** https://youtu.be/3_LkyVcPHiU?si=qDRI8wDErUH_oKoD
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5302

When an exploit starts working in production, the gap between "we are leaking data" and "a reviewed fix is deployed" is measured in hours, and the attack keeps working the whole time. Automating it is hard for two reasons: a firewall rule is not a fix (attackers change strings), and a model-written patch cannot be trusted blindly.

OODA-Aegis is an autonomous defense agent that closes that gap on Meta's open-weight models, without a human and without data leaving the network. It runs an Observe, Orient, Decide, Act loop on two horizons:

## More from Meta Global AI Developer Hackathon

- [Quien Tiene (Axon)](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5299)
- [Deskverse](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5300)
- [KushZombie](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5301)
- [ERPClaw by AvanSaber](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5303)
- [Ride AI](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5304)
- [Easy Onboarding](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5305)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5302. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt