Source Boundaries
Guida currently ships as a closed-source Windows app. That does not mean every part of the project is meant to stay private forever, and it also does not mean the whole app is moving to open source.
The boundary is practical: reusable infrastructure can be useful outside Guida, while the product-specific browser automation and web scraping layer remains part of the closed product.
Closed product layer
Section titled “Closed product layer”The browser automation and web scraping implementation will stay closed source.
This includes the product-specific layer that coordinates browser control, extraction behavior, scraping workflows, and the guardrails around those capabilities. It is both core to Guida as a product and the area where redistributed builds with removed safeguards would carry the highest risk.
App shell
Section titled “App shell”Guida is a Windows desktop app built with WPF and WebView2. The app shell coordinates panes, browser tabs, workspaces, script execution, MCP surfaces, and local persistence.
The app currently ships as a binary, not as an open-source repository.
Scripting layer
Section titled “Scripting layer”Guida scripts use the g.* API from JavaScript, Lua, and Janet. This layer includes runtime hosting, API registration, script context creation, editor metadata, and language-specific bindings.
The scripting engine and related infrastructure will be open-sourced once the boundary is extracted cleanly and the code is documented enough to be useful outside the app.
MCP and security layer
Section titled “MCP and security layer”Guida includes an embedded MCP server with tools, resources, prompts, approval checks, audit logging, origin tracking, and secret isolation.
The reusable parts of secrets management, auditing, and related security infrastructure will be open-sourced once they can stand on their own. The browser automation implementation that those surfaces can call remains part of the closed product layer.
Workspace layer
Section titled “Workspace layer”Workspaces provide local project state: scripts, views, workflows, queues, store data, search indexes, event triggers, and configuration files.
Workspace-related infrastructure will only be open-sourced where it can stand on its own without exposing product-specific scraping behavior.
Already public
Section titled “Already public”Two packages created while building Guida are already available:
- JanetSharp — Janet runtime integration for .NET.
- StratQueue — SQLite-backed work queues with pluggable dequeue strategies.
Future direction
Section titled “Future direction”The open-source direction is reusable infrastructure: the scripting engine, secrets management, auditing, and related support code.
I am not publishing a date yet because the extraction boundary still has to be made clean. The important part is the direction: share infrastructure that is useful beyond Guida, keep the product-specific browser automation and web scraping layer closed.