iterative-code-exploration

Use an iterative loop: discover entry points and structure with find and tree, evaluate file relevance by reading READMEs and package manifests, refine by tracing specific modules and imports, then repeat up to 3-4 cycles until you can explain the main flows.

Search for definitions with patterns like 'function name' or 'const X =', and find usages with grep -r "functionName(" across the source directory. For imports, search 'import.*X' to trace dependency direction.

Useful patterns include grep -r for class and interface definitions, searching 'app.get' or '/api/' for endpoints, 'process.env' for environment variables, and 'TODO|FIXME' for known problem areas. Combine with find to locate entry points like main.* or index.* files.

Stop when you can explain the core logic, trace the main data flows, and feel confident making changes. The method caps exploration at 3-4 loop iterations; if key functionality remains unclear after that, narrow your scope to one feature instead of the whole system.

Yes, but scope each loop to a single package or feature rather than the entire monorepo. Start with top-level configs and workspace manifests to map package boundaries, then apply middle-out feature tracing within the relevant package.