Author

Luz Wang

Founder and engineer, Terse.

About

Luz Wang builds Terse — an on-device optimizer for AI coding agents that compresses prompts before they reach the API, monitors eight agents live, and caps runaway spend with a budget circuit breaker.

The work on this site comes out of building that: the guides here are about where tokens actually go inside an agent loop, which of them are avoidable, and what the compression is worth once context carries forward turn after turn.

Verify

GitHub
github.com/lucaszengool — the account that authors Terse's commits
Project
github.com/Terse-AI/terseai — releases and changelog
Contact
[email protected]

Writes about

token optimization prompt compression AI coding agents Claude Code Cursor MCP prompt caching context windows Rust Tauri

Selected writing

← All writing on Terse