PROTOCOL: AEGIS

AI defense layer for decision environments

AEGIS is not another adblocker. It reads the web as a decision environment and detects pressure patterns such as urgency, scarcity, confirmshaming, asymmetric cookie banners, and aggressive tracking layers.

What AEGIS solves

• Detection of manipulative patterns across text, visual cues, and page behavior
• Rewrite of pressure copy into a more neutral form
• Suppression of urgency effects without breaking layout
• Cookie banner normalization for a clearer consent flow
• DOM revert for reversible interventions
• Auditable intervention model for future team and enterprise modes

MVP technical loop

The current stack combines a product website, browser extension runtime, local inference bridge, and shared typed contracts. Runtime works in a closed loop:

scan → classify → rewrite / suppress → audit → revert

The extension layer is built on Manifest V3. It includes a content scanner, background orchestration, sidepanel UI, inference bridge, and /api/aegis/evaluate endpoint. If Ollama is running, the system can use a local model; otherwise it falls back to a local policy engine.

Product positioning

AEGIS defines a cognitive defense software category. It protects not only against ads and tracking, but against manipulation of the decision environment itself.

Roadmap

• End-to-end verification of unpacked Chrome flow
• Live threat feed based on runtime data
• Audit trail export
• Dataset of manipulative patterns in Czech and other languages
• Team licensing and enterprise on-prem policy mode

The project runs at aegis.d-international.eu.