Code agents need better search than grep alone
can miss the real working function when they rely only on grep-style text search. Letting the agent read whole files to investigate also uses up the quickly. A setup can work better because it combines exact word matching with meaning-based search.
Adding helps sort the first results again so the agent sees the most useful code sooner. This can reduce wasted reading and help the agent reach the right part of the codebase faster.
Key points
- grep alone can miss important code when the wording does not match.
- Reading whole files can quickly use up the .
- combines text matching with meaning-based lookup.
- helps put the most useful code results first.
- Better code can lower token waste in .