Brave Core
repository·master·Indexed 25 days ago
https://github.com/brave/brave-coreA specialized set of changes, APIs, and scripts used to customize the Chromium source code to produce the Brave browser for Android, iOS, Linux, macOS, and Windows. It manages Chromium syncing, Brave-specific patches, and dependencies like adblock-rust. Documentation covers internal architectures including the Brave NTP React application, local AI and history embeddings using EmbeddingGemma, the Brave VPN WireGuard Service for Windows, and the Endpoint Client for type-safe API requests.
What's inside brave-core
- A 'Study' in Brave Core is a collection of experiments conducted directly on the client. These studies are used to manage and execute various feature variations and experiments. For more detailed information on the underlying mechanism, refer to the Brave Variations (Griffin) documentation.
Overview of Plaster semantical patching
masterPlaster is an experimental tool used to apply changes to upstream source files (specifically Chromium'ssrcrepo) using regex transformations or AST-based rewriters. Unlike traditionalgitpatches, which are brittle during rebases, or language overrides, which can be hard to audit, Plaster provides a language-agnostic, semantic way to apply changes in place. It generates standard patch files as a side effect, allowing changes to be audited and integrated into the existing Brave patch machinery.Overview of Brave VPN WireGuard Service
masterThe Brave VPN WireGuard Service is a Windows service designed to manage WireGuard VPN connections. It provides both COM (Component Object Model) and CLI (Command Line Interface) interfaces, allowing users or other applications to start and stop a WireGuard VPN connection by providing a specific configuration file.Overview of anonymous-credentials
masterThe
anonymous-credentialslibrary is a partial implementation of Direct Anonymous Attestation (DAA) designed for the Web Discovery Project. It provides signer functions that perform elliptic curve operations.Supported Curves:
bn254is the only supported curve for this library.
Overview of Reactions in Brave Ads
masterThe Reactions system allows users to provide feedback on specific segments, advertisers, or individual ads. This feedback mechanism is used to influence the types of ads a user is shown in the future.Overview of Condition Matchers
masterCondition matchers are used to evaluate ad targeting criteria. They function by comparing specific targeting criteria against defined operators and patterns to determine if the criteria are met.Overview of Condition Matcher for ad targeting
masterTheCondition Matcheris a component used to assess whether specific conditions or criteria are fulfilled for the purpose of ad targeting. It serves as the logic layer to determine if an ad should be served based on predefined requirements.Overview of Geographical Targeting
masterGeographical Targeting allows for the delivery of advertising content within defined geographic boundaries to users who meet specific criteria.Overview of Brave Ads Core
masterThebrave_ads/corecomponent provides a platform-independent implementation of the Brave Ads platform. It is designed as a privacy-preserving global ad platform. For specific terminology and domain concepts used within this implementation, refer to the project's glossary of terms.Overview of Refill Confirmation Tokens URL Requests
masterRefill confirmation token URL requests are used to fetch signed tokens specifically for Brave Rewards users. These tokens are part of the mechanism for confirming account refills.Overview of Brave SpeedReader
masterBrave SpeedReader is a tool designed to distill text-focused content from HTML documents. It operates on HTML before rendering, producing a clean HTML output that uses Brave-designed themes and contains no external styling or scripting.
Key features include:
- Distills text-focused content from suitable HTML.
- Works on raw HTML before the rendering process.
- Generates lightweight HTML output without external dependencies.
Overview of Text Classification Targeting
masterText Classification Targeting is a machine learning technique used to target ads by predicting predefined taxonomy segments from text content. This allows for ad placement based on the semantic meaning of the text being processed.