$ loading_
LLM-native MCP tools for reading, understanding, and safely modifying code
Copy the install command and let the AI configure it · recommended for beginners
No copy-paste install info for "LiuHe" yet — see the docs or source repo.
Use LiuHe to scan this repository and summarize its main modules, symbol index, and key dependencies so I can quickly understand the codebase.
A repository overview, important symbol relationships, and a code map for further analysis.
Use LiuHe to refactor this module without breaking the existing structure, and apply deterministic quality checks before writing any changes. Keep only changes that pass.
Validated code changes, plus which edits passed or failed the quality gates.
Use LiuHe to analyze code files across multiple directories concurrently and identify reusable functions, repeated patterns, and areas needing closer review.
Analysis results across directories, including reuse opportunities, duplicate-code signals, and recommended review areas.
Developers can use it when AI agents need to read, understand, and modify large codebases. Its transactional writes and quality gates help reduce the risk of bad edits.
When teams need to locate symbols and relationships quickly, this tool uses a SQLite symbol index and repository map to improve understanding speed. It fits code navigation and structural analysis workflows.
In resource-constrained or high-concurrency agent environments, it emphasizes crash-isolated parsing and low memory usage. It suits stable, large-scale code processing tasks.
It is an MCP toolset for AI agents, offering 44 code tools to read, understand, and modify code. The description also mentions symbol indexing, transactional writes, and deterministic quality gates.
According to the description, its deterministic quality gates use zero LLM calls. That means this part of validation does not depend on model calls, which can make checks more stable and repeatable.
The provided material does not include installation steps, runtime requirements, or key requirements. For integration details, see the source repository.
Parse repositories, index symbols, and answer codebase questions autonomously.
Gives AI coding agents a structural map of your repository fast.
Give AI coding agents filesystem, Git, database, and compute tools via MCP.
Retrieve GitHub code at symbol level to explore repos with far lower AI token costs.
Analyze code structure and Git history while drastically reducing AI token usage.
Search and navigate multiple code repositories with natural language understanding.