How does ogrep differ from traditional grep?

While grep searches for literal text strings, ogrep uses semantic embeddings to understand the meaning of your query, finding relevant code even if the keywords don't match exactly.

How can I reduce token usage when searching large repos?

You can use the --summarize flag to get a file-level overview first, allowing you to narrow down the relevant files before drilling into specific code chunks.

Does ogrep support multi-branch development?

Yes, ogrep is branch-aware. It tracks files per git branch to ensure that your search results are always relevant to the code you are currently working on.

What is AST-aware chunking?

AST-aware chunking uses tree-sitter to understand code structure, ensuring that search results return complete, logically coherent blocks (like full functions or classes) rather than arbitrary line ranges.

ogrep: Semantic Code Search

Name: ogrep: Semantic Code Search
Author: gplv2

bygplv2

•

开发者工具

Performs semantic search to find code by meaning and intent rather than relying on literal keyword matches.

关于

ogrep is a specialized search utility for Claude Code designed to solve the 'cold start' problem in unfamiliar codebases. By leveraging semantic embeddings, it allows developers to locate logic and implementation details through conceptual queries—such as asking 'how are payments handled'—even when exact function or class names are unknown. The skill features AST-aware chunking for structurally coherent results, branch tracking to ensure search accuracy across git history, and a robust core loop that helps Claude navigate, explore, and extract precise code evidence with line-level citations.

主要功能

Confidence scoring and reranking to prioritize the most relevant results
3 GitHub stars
Branch-aware indexing to maintain search integrity across git branches
Semantic, hybrid, and full-text search modes for high-precision discovery
AST-aware chunking for Python, JavaScript, TypeScript, Go, and Rust
Context expansion (chunks) to view surrounding code without file dumping

使用场景

Finding specific code implementations based on behavioral descriptions
Locating cross-cutting concerns like error handling or auth logic
Navigating and understanding large or unfamiliar codebases quickly

关于

主要功能

Confidence scoring and reranking to prioritize the most relevant results
3 GitHub stars
Branch-aware indexing to maintain search integrity across git branches
Semantic, hybrid, and full-text search modes for high-precision discovery
AST-aware chunking for Python, JavaScript, TypeScript, Go, and Rust
Context expansion (chunks) to view surrounding code without file dumping

使用场景

Finding specific code implementations based on behavioral descriptions
Locating cross-cutting concerns like error handling or auth logic
Navigating and understanding large or unfamiliar codebases quickly