unified-search-discover

Progressive search with safe discovery enforces a COUNT → PREVIEW → EXECUTE lifecycle. Start by counting matches with fd, rg, or ast-grep to gauge scope, preview samples to validate correctness, then execute the full refactor or replacement with backup guidance to prevent unintended changes.

File discovery with fd or find respects gitignore rules by default, filtering out ignored paths automatically. This portable approach works across Unix and Windows environments, minimizing noise and focusing results on tracked source files.

Yes, ast-grep enables AST-based search and replace by matching code structure rather than text patterns. This catches semantically equivalent code that text regex would miss and reduces false positives when refactoring.

rg (ripgrep) performs fast text searches across files with built-in gitignore support and parallel processing. Use it in the COUNT phase to quickly filter results before previewing and executing changes.

The COUNT → PREVIEW → EXECUTE pattern minimizes risk by validating scope and samples before full execution. Always maintain backups and review preview output to ensure the refactor matches your intent before committing changes.

Yes, fd and find with gitignore awareness adapt to varied project layouts while respecting version control rules. The portable design works consistently across TypeScript, JavaScript, and mixed-language projects.