From prompt to Excel custom function in 30 seconds (alexjreid.dev)

xllify is a packaging tool and runtime that exposes Luau and Python functions as high-performance Excel custom functions by producing .xll extensions via the Excel C SDK. The author wired a Claude agent into xllify’s command-line tooling so a single natural-language prompt produced code, exercised existing CLI tests, and built a ready-to-load .xll — all in about 30 seconds. The demo produced a working custom formula (=VictorianCompliment) with no manual coding, illustrating how an LLM agent can orchestrate compilation, testing, and packaging steps in a UNIX-style pipeline. For the AI/ML community this matters because it shows LLMs moving beyond prose/code generation into automated toolchain orchestration: agents can integrate legacy systems (Excel’s C SDK), domain languages (Luau), and Python runtimes to rapidly create deployable artifacts. That lowers barriers to extension development and unlocks productivity gains, but raises quality, security, and maintainability concerns — generated artifacts still need validation and human oversight. The experiment highlights a practical pattern for LLM-driven developer tooling: combine small, testable utilities (pipes) with agent orchestration to turn prompts into runnable binaries, accelerating iteration while underscoring the need for robust verification and engineering maturity.