Show HN: Sweet-search – local code search for agents (5–33% cheaper runs)

Category: devtools

Tags: code-search, ai-agents, local-inference, semantic-search, developer-tools, claude-code

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

Sweet-search is a local code search and indexing tool designed specifically for AI coding agents like Claude Code and Codex, offering six specialized tools (search, grep, find, semantic, trace, read) to replace naive grep+Read workflows. It builds a GPU-accelerated local index with incremental updates and provides benchmark claims of 5–33% cheaper agent runs. The project integrates directly into agent harnesses via system prompt injection, MCP, and CLI, emphasizing 100% local inference and privacy.

Target audience: AI developers, backend engineers, DevOps engineers using coding agents

Repository: https://github.com/mrsladoje/sweet-search · JavaScript · Apache-2.0 · 53 stars

View on Hacker News