wulun811/LiuHe

Overview

LLM-native code toolkit combining a Rust multi-language parser (tree-sitter) with 44 MCP tools for atomic editing, impact analysis, reference tracing, and deterministic zero-LLM code quality gates. Designed for LLM agents operating without direct human interaction, providing structured code access and modification.

Best for

Best for
Developers building MCP-based coding agents that need deterministic code manipulation.

Use cases

• Apply atomic code edits through MCP tools
• Trace symbol references across languages
• Run deterministic quality gates without LLM inference

Notes

LLM-native code toolkit combining a Rust multi-language parser (tree-sitter) with 44 MCP tools for atomic editing, impact analysis, reference tracing, and deterministic zero-LLM code quality gates. Designed for LLM agents operating without direct human interaction, providing structured code access and modification.

2 stars on GitHub. Last updated 2026-09-03. Licensed MIT.

Use cases

• Apply atomic code edits through MCP tools
• Trace symbol references across languages
• Run deterministic quality gates without LLM inference

Pros

• Multi-language support via tree-sitter
• 44 MCP tools cover editing, analysis, and tracing
• Zero-LLM quality gates produce consistent results

Cons

• Very low adoption (2 GitHub stars)
• Requires MCP-compatible environment
• Limited community validation or documentation

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.