Home
Contact
← Back to HomeView on GitHub

Chrome Page Sender

Motivation

I wanted a way to let local AI powered tools see the exact page I was looking at without building a full browser automation stack. This extension bridges that gap by capturing the active tab and sending it to a local endpoint that I control.

It started as part of a job application workflow, where an agent could compare a posting against my resume and project history. It also helped friends who track sports betting lines, replacing manual copy and paste with a single click while still respecting site rules against scraping.

Those are just two examples. The goal was to keep the tool flexible enough to adapt to a wide range of workflows.

Technical Overview

  • Manifest V3 service worker: A lightweight background script coordinates capture requests and handles network communication.
  • Tab capture pipeline: Uses Chrome Extension APIs to grab the active tab title, URL, and text content, with optional HTML capture for richer context.
  • HTTP click to send: Clicking the extension icon posts a JSON payload to a local `POST /ingest` endpoint for immediate ingestion.
  • WebSocket bridge: A local client can request capture or tab lists on demand, which makes it easy to integrate with Electron or other desktop tooling.

Design Choices & Tradeoffs

  • Local first architecture: Everything routes to localhost so sensitive pages never leave my machine unless I choose to forward them.
  • Optional HTML capture: Full HTML can be useful for structured parsing, but it is disabled by default to keep payloads light.
  • User driven capture: Data is only sent on click or explicit WebSocket request, avoiding passive scraping and keeping the behavior transparent.

Source & License