colgrep

Search code repositories by intent with semantic retrieval and regex filtering.

527|61|Updated Jan 6, 2026
One-click install
npx skills add https://github.com/lightonai/next-plaid --skill colgrep-lightonai
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: colgrep
Source: https://github.com/lightonai/next-plaid/tree/main/colgrep/src/install
Command: npx skills add https://github.com/lightonai/next-plaid --skill colgrep-lightonai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Colgrep helps you locate the right code faster when exact filenames, symbols, or literal text are not enough. It combines semantic search with regex filtering so you can discover relevant code paths by intent.

Core Features & Use Cases

  • Semantic code search: Search repositories using natural language queries like a concept, behavior, or bug description.
  • Hybrid filtering: Combine meaning-based ranking with regex, glob-style includes, excludes, and file scoping for precise results.
  • AI tool integration: Install the skill across coding assistants such as Claude Code, Codex, OpenCode, Hermes, and Kimi Code so they default to colgrep for code search.
  • Use case: Ask for the code that handles database connection pooling, error handling, or async workflows, and get ranked matches without manually guessing search terms.

Quick Start

Ask your coding assistant to use colgrep to find the code that implements the behavior you are looking for.

Frequently Asked Questions about colgrep

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I search a code repository by meaning instead of exact text?

Semantic code search lets you find code by meaning using natural language queries. Colgrep combines meaning-based ranking with regex filtering to locate relevant code paths by intent rather than exact text matches.

How do I find code related to a specific behavior like database connection pooling?

You can find code handling database connection pooling by asking your coding assistant to use colgrep. It applies semantic retrieval to rank matches based on your behavioral description without guessing exact search terms.

Can I combine natural language search with regex filtering for code exploration?

Hybrid filtering combines meaning-based ranking with regex, glob-style includes, excludes, and file scoping. This lets you apply precise text constraints alongside semantic code search for accurate repository exploration.

Does semantic code search work with AI coding agents like Claude Code or Codex?

Semantic code search integrates with AI coding assistants including Claude Code, Codex, OpenCode, Hermes, and Kimi Code. Colgrep installs as a skill so these agents default to semantic retrieval for code discovery tasks.

What is the best way to explore an unfamiliar codebase and understand its system architecture?

Repository exploration using semantic search helps you understand unfamiliar systems by locating code paths through behavioral queries. Colgrep replaces brittle text-only search with intent-based ranking for bug hunting and refactoring.

Why does standard text search fail when hunting for bugs across a large repository?

Text search fails during bug hunting when exact symbols or filenames are unknown. Semantic code search solves this by matching your bug description to relevant code paths through intent-based ranking and hybrid filtering.