Show HN: Semble – Code search for agents that uses 98% fewer tokens than grep

Category: devtools

Tags: code-search, agent-tools, token-efficiency

Score: 7.0/10 (Innovation: 6, Technical: 7, Documentation: 8, Utility: 7)

Semble is a code search library designed for AI agents that returns relevant code snippets using dramatically fewer tokens than traditional grep+read workflows. Its combination of fast indexing on CPU, natural language queries, and MCP/Bash integration makes it a practical tool for improving agent efficiency in large codebases.

Target audience: backend devs, devops, ai-engineers

Repository: https://github.com/MinishLab/semble · Python · MIT · 830 stars

View on Hacker News