semble

Index and search code across local and remote repositories with token-efficient semantic queries.

40|9|Updated Jan 4, 2026
One-click install
npx skills add https://github.com/akillness/jeo-skills --skill semble
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: semble
Source: https://github.com/akillness/jeo-skills/tree/main/.agent-skills/semble
Command: npx skills add https://github.com/akillness/jeo-skills --skill semble

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Semble provides token-efficient, semantic code search that returns only relevant code chunks for AI agents, avoiding full-file grep overhead.

Core Features & Use Cases

  • Natural-language and symbol code search across local or remote repositories.
  • Semantic similarity search and find-related capabilities to discover related code patterns.
  • MCP integration across Claude Code, Codex, Cursor, and OpenCode, plus a Python library for programmatic access.

Quick Start

Install semble, index your project, and perform a natural-language search to locate relevant code.

Frequently Asked Questions about semble

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

FAQPage Schema
How do I perform semantic code search across local and remote repositories for an AI agent?

Semantic code search across local and remote repositories is achieved by indexing code chunks and querying them with natural language or symbols, returning only the most relevant matches to avoid full-file grep overhead for AI agents.

Does token-efficient code search require a GPU or API keys to run?

Token-efficient code search requires zero GPU or API keys to run. It achieves high-speed indexing at around 250ms and fast queries at approximately 1.5ms without external dependencies or specialized hardware.

Can I integrate semantic code search with Claude Code, Cursor, and OpenCode?

Semantic code search integrates with Claude Code, Cursor, and OpenCode through MCP server integration. It also provides a Python library for programmatic access to query indexed repositories.

What is the best way to discover related code patterns without reading entire files?

Discovering related code patterns without reading entire files is done using semantic similarity search and find-related capabilities, which locate and return only the specific relevant code chunks.

How do I query code using natural language and symbols in a large project?

Querying code using natural language and symbols in a large project involves indexing the repository first, then executing semantic or programmatic searches that quickly return targeted code chunks.

What are the limitations of using grep for AI agent code discovery versus semantic search?

Grep limitations for AI agent code discovery include full-file overhead and lack of semantic understanding. Semantic search overcomes this by indexing and returning only relevant code chunks using natural-language queries.