yuxino/kiri

A fast, local-first workspace for screenshots, annotation, OCR, and screen recording.

What it solves

Kiri is a local-first screen capture tool that combines screenshots, annotations, OCR, and region recording into a single application. It eliminates the need for multiple separate tools for capturing images, extracting text from images, and creating short videos or GIFs of your screen.

How it works

Kiri uses system-level APIs (macOS Vision and Windows.Media.Ocr) to perform text recognition locally on your device. It captures screen regions or windows, providing a built-in editor for adding shapes, arrows, and mosaics. For recording, it leverages AVFoundation (macOS) and Media Foundation (Windows) to produce MP4s and GIFs without requiring external dependencies like FFmpeg.

Who it’s for

Users on macOS and Windows who need a fast, productivity-oriented capture tool that prioritizes privacy by keeping data and processing local by default.

Highlights

  • Local OCR: Recognizes text locally using native system APIs, with an optional remote OCR feature.
  • Local Library: A built-in manager to search, tag, and favorite your captures.
  • Integrated Recording: Captures screen regions as MP4 or GIF with optional system audio and microphone support.
  • Annotation Tools: Includes a full suite of editing tools like crop, pen, shapes, and mosaic for quick visual communication.

Related

  • Project
    paperclipinc/openclaw-operator

    A Kubernetes operator for self-hosting OpenClaw AI agents, providing production-grade security, observability, and automated lifecycle management through a declarative API.

  • Project
    mudler/vllm.cpp

    A lightweight C++20 inference engine that mirrors vLLM's features and throughput but removes the Python/PyTorch dependency for a 140x smaller installation footprint.

  • Project
    Entrpi/ds4-on-spark

    A high-performance serving stack for DeepSeek-V4-Flash on NVIDIA DGX Spark, featuring continuous batching, speculative decoding, and a memory governor for massive context windows.

  • Project
    NVIDIA/makani

    A PyTorch library for the massively parallel training and inference of machine-learning weather and climate models, supporting scaling across 100+ GPUs.