Inspiration
TypeMind draws on the fluid in-editor suggestions of tools like Cursor and GitHub Copilot, aiming to deliver that same focus-preserving flow directly inside Chrome while keeping user data local.
What I Learned
I became proficient with Chrome AI’s Prompt API, including session lifecycle management and guardrail handling.
How I Built It
I collaborated closely with the Codex CLI to iterate on implementation plans, validate the architecture, and assemble the build. The extension is split into a Vite-powered background service worker, a React + Tailwind popup, and an IIFE content script that injects a Shadow DOM overlay into editable surfaces—all coordinated through a session-aware completion engine under src/core.
Challenges Faced
I optimized prompts to shorten the path from keystroke to suggestion, avoiding the slowdown caused by sending excessive context to the Prompt API. By reusing Prompt API sessions, I balanced freshness with responsiveness for sustained performance gains. Supporting both textareas and rich contenteditable fields required carefully tracking selection state so users’ text was never overwritten.
Log in or sign up for Devpost to join the conversation.