own product · Live, 232 users
Screencap
Screencap is a Chrome extension by Haris Muhibić for capturing, annotating and sharing screenshots, with on-device OCR. It has 232 users and a 5-star rating.
- Role
- Founder, sole engineer
- When
- February 2026 – present
- Links
- screencap.app
- Stack
What it is
Capture any part of a page, annotate it, make it presentable and share a link. Text recognition runs on your device, so screenshots never leave the browser just to read text from them.
What I built
- A Manifest V3 extension in six languages.
- A FastAPI and Postgres backend for share links, with Google sign-in and separate API and file domains.
- An abuse policy with image moderation, plus Prometheus and Grafana monitoring.
- A landing site, blog and changelog on Cloudflare Workers.
Engineering details
- OCR bundle from about 54 MB to 13 MB: I ship only the LSTM assets and pick the relaxed-SIMD, SIMD or plain WASM core at runtime.
- Faster results: the OCR worker warms up while you are still reviewing the capture, and a second pass runs only when the first result has low confidence.
- Agent access by design: a written architecture for a screen-recorder MCP server. Agents control recording through a native companion, and media bytes, file paths and tokens never cross the MCP boundary.
Why it matters for clients
It shows what happens after launch: store review, users, moderation, monitoring and steady releases.
Building something similar? Book a 30-minute call or email haris@chern0.dev.