Skip to main content

Module browser

computer_use

Module browser 

Source
Expand description

P10: Chrome DevTools Protocol (CDP) attach-only client.

The CLI thread stays sync. Every browser call spins up a fresh tokio::runtime::Builder::new_current_thread() runtime, runs the async CDP request, and drops the runtime before returning. The backend never spawns Chrome — the user already started the browser with --remote-debugging-port=N, and we just attach.

Built only when the cdp feature is on. The default build exposes the unsupported error code and nothing else.

§Port discovery

When the caller passes --cdp without --cdp-url, we look up the Chrome pid that owns the --app (or the frontmost one if no app is named) and read its argv:

  • macOS: sysctl KERN_PROCARGS2 for the pid.
  • Linux: /proc/<pid>/cmdline (NUL-separated UTF-8).
  • Windows: best-effort; not implemented in this phase (see ROADMAP.md).

Only the first matching --remote-debugging-port=N wins. The handler ws://127.0.0.1:<N> is what we hand to chromiumoxide::Browser::connect.

§Viewport -> screen mapping (P5 reuse)

browser_click resolves a CSS selector, asks the element to click, and chromiumoxide clicks in CSS pixel space. The screenshot layer maps those coordinates into physical screen coordinates for the existing P3 click path. HiDPI and CSS pixels-above-fold cases still rely on the configured coord ratio like the rest of the codebase; we do not introduce a new viewport transform here.

Re-exports§

pub use cdp::discover_debug_port;
pub use cdp::parse_debug_port;
pub use cdp::ActionOptions;
pub use cdp::AttachOptions;
pub use cdp::TabSummary;

Modules§

cdp
P10: Chrome DevTools Protocol (CDP) client.