Our summary
jgrep searches for meaning rather than only matching exact words. It asks the same yes-or-no question about each chosen piece of text, such as a line, paragraph, document, or code change. Matches retain their original wording and location so a reader can investigate them.
The guide explains how to include nearby context, reuse saved answers, limit spending, and estimate work before sending requests. When reviewing a code change, it keeps removed and added lines together. Optional extra context can include the surrounding function, a named block of code.
A match estimate does not explain why a passage matched, and missing context can change the answer. Extra code context works only for supported languages. Source text may leave the computer, while reported time and cost depend on the service and material used.
Key takeaways
- Choose a piece of text large enough to preserve its meaning.
- Estimate calls and cost before a large search.
- Read the original matches instead of treating their scores as explanations.