Parsers are of course the foundation of any language tooling, because your tools need to know what some source code actually says. Most of Working Effectively with Legacy Code deals with how to bootstrap your way to safely making changes in a codebase that lacks tests, when writing tests requires changing the codebase to make it testable. The TLA⁺ Foundation has been funding me for this in various ways, such as writing developing & contributing guides, highlighting good first issues in monthly development updates, and - most substantially - writing a long tutorial on how to write your own minimal TLA⁺ model checker!