Researchtrigramsregex searchcode searchcursor
Cursor Builds Trigram Indexes For Regex Search
8.1
Cursor is building trigram-based inverted indexes to speed regular-expression searches for Agents across large monorepos, citing ripgrep's 15+ second scans in enterprise repos. The approach adapts decades-old trigram decomposition and regex tokenization to prune candidate files before full matching, improving interactive agent responsiveness and reducing latency when guiding code generation.
Key Points
- 1Indexing: Cursor implements trigram inverted indexes to accelerate regular-expression searches over large codebases.
- 2Performance: ripgrep matches file contents fast but scans all files, causing 15+ second searches in monorepos.
- 3Implication: Trigram decomposition lets agents prune candidate files before regex matching, improving responsiveness and workflow.
Scoring Rationale
Practical engineering application with measurable impact on agent workflows; limited novelty because trigram indexing is a known technique.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
