Semble - grep보다 토큰을 98% 적게 쓰는 에이전트용 코드 검색

Semble solves the token cost problem that has limited AI agent adoption in code exploration. By using semantic code chunking instead of returning entire files, it cuts token consumption by 98% compared to grep while maintaining search quality. Local execution and direct MCP integration with Claude Code and Cursor make it immediately practical for developers building LLM-powered workflows.