Content Filtering

Quira includes a high-performance content filtering engine built directly into the browser's networking pipeline. It blocks unwanted content — ads, trackers, cryptominers, and malicious scripts — before they load, saving bandwidth and protecting privacy.

Filter engine

The content filter engine uses a Rust-based implementation that evaluates filter rules against network requests in constant time using compiled Aho-Corasick automata. This approach is significantly faster than regex-based engines used by traditional browser extensions.

• Network-level blocking — Requests are blocked before any data is downloaded
• Cosmetic filtering — CSS-based element hiding removes visual remnants
• Scriptlet injection — Neutralizes anti-adblock scripts safely
• CNAME uncloaking — Resolves CNAME records to detect disguised trackers

Built-in filter lists

Script sandboxing

Third-party scripts are evaluated in a restricted sandbox that limits their access to browser APIs. All script execution events are monitored by the Context Security Event Bus (CSEB), and anti-fingerprinting behavior is further strengthened by the Adaptive Graph Topology Randomization (AGTR) technique. Scripts from untrusted origins cannot:

• Access the clipboard or file system
• Open popups or navigate the main frame
• Access cross-origin storage
• Use fingerprinting APIs (Canvas, WebGL readback, AudioContext) without user consent

Cosmetic filtering

After network-level blocking removes the bulk of unwanted content, cosmetic rules hide any remaining visual artifacts. Quira applies CSS display: none rules and uses a MutationObserver to catch dynamically inserted elements. Cosmetic rules are page-specific and do not degrade rendering performance.