Automating everything between the prompt and the commit. I build the tools AI agents use: SQL engines over dev data, stateless debuggers, visual feedback loops, and execution pipelines that turn intent into shipped code.
Query AI coding history joined with Git commits
SQL engine that unifies Claude Code conversation history with Git repository data. Join AI interactions with commits, diffs, and branches to analyze development patterns and productivity.
Terminal-aesthetic React component library
54-component React library with MS-DOS/terminal aesthetics built on Base UI. Monospace typography, box-drawing borders, and full theming via CSS classes. This site is built with it.
Dual-interface text editor (TUI + GUI)
Text editor with both a terminal UI and graphical interface sharing the same engine. Built in Rust with a focus on keyboard-driven workflows and extensibility.
MCP server for AI-driven Rust debugging
Model Context Protocol server that gives AI agents deep access to Rust project analysis. Provides structured debugging context, crate inspection, and compilation diagnostics to LLMs.
18M ops/sec schema validation in Rust/WASM
High-performance schema validation library for Rust with WASM bindings. Achieves 18 million operations per second for common validation patterns.
SQL over Claude Code conversation data
Query your Claude Code history using SQL. Analyze conversations, tool usage patterns, and token consumption across your AI-assisted development sessions.
Stateless CLI debugger for Node.js + browsers
Debug Node.js processes and automate browser pages using a stateless CLI interface. Designed for AI agent workflows where traditional debuggers break down.
Visual feedback via screenshot annotations
CLI tool for annotating screenshots with circles, arrows, and text. Built for AI agent workflows where visual feedback from humans guides automated actions.