Asked to explain how a checkout request is processed in an unfamiliar 600-file service, an agent starts by reading every file in src/. It runs low on context before reaching the payment code. What approach should it use instead?

Answer: C — Grep for the checkout route handler, Read it, then follow its imports one module at a time

Incremental exploration starts with Grep to find the entry point, then uses Read to follow imports and trace the flow, reading only files on the relevant path (c). File-name matching (a) is a guess that misses code without 'checkout' in the path and still reads files upfront. Summarizing every file (b) and reading one concatenated dump (d) both still load the whole codebase.