Osprey ports the pydantic-ai 2.46.0 agent engine to its Python flavor and builds Osprey Agent, a desktop coding agent, on that port and Osprey Native.
Badges
Badges come from GitHub once the author claims this feed.
Recent posts
every link opens the author’s post-
Osprey Agent: a Coding Agent on the pydantic-ai Port
-
An Accessibility Tree AI Agents Can Drive: Automated UI Testing by Label
What an accessibility tree is, why AI agents and automated UI tests work better from one than from screenshots, and how Osprey's UI machines and @osprey/uitest drive a native app by label with no window, shown on a Chromecast player.
-
How to Chromecast From a Mac to a TV: A Native Player Written in Osprey
Cast music from a Mac to a Chromecast or Google TV without Chrome or a phone: find the TV, open an MP3, play, pause, seek, change volume and speed, and show a spectrum visualiser on screen.
-
Static Typing for Python-Style Code, Without Writing Type Hints
What static typing is, how it differs from dynamic typing, and how Osprey's Python flavor checks Python-style code at compile time with inferred types instead of hints, shown on a Chromecast sender ported from a CPython script.
-
A Rust UI Framework Without Writing Rust: GPUI Apps as State Machines
What GPUI is, how it compares with egui, Slint and Tauri, and how Osprey's UI machines describe a native GPUI app as state machines and skins in one JSON document, shown with a Chromecast player.
-
Desktop Apps on GPUI, Written in Osprey
How an Osprey application runs in a GPUI window through Osprey Native: the C boundary, scenes and paths as data, the frame clock, measured frame rates, and how this differs from Osprey's GPU compute surface.
-
Python Packages in Osprey: NumPy, PyTorch-Style Code, and uv
Understand Osprey's Python-flavor packages, NumPy and PyTorch-style ports, and the new package manager's verified source locks and uv integration for CPython dependencies.
-
Automated Testing with Osprey Hunt: Find the Input That Crashes
Use symbolic execution for automated testing in Osprey: reproduce a crash, read HTML and JSON reports, and understand where Hunt fits beside static analysis and fuzzing.