Mantic.sh MCP
A structural code search MCP server for agents: finds the right file by query intent and runs locally
- 553 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Reads and indexes your code locally during search
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/mantic-shDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Reads and indexes your code locally during search
This is third-party code. Review the repository files before installing.
What it does
Mantic is a code search engine that bets on result relevance rather than raw speed. It infers intent from project structure and metadata, filtering out irrelevant files before reading content, which cuts tokens spent on noise. It offers go-to-definition and finding all uses of a symbol across a monorepo while respecting gitignore. Optional semantic reranking on local embeddings finds conceptually close code without an exact keyword match. It works as a CLI and as an MCP server for agents, all locally and with no data egress.
Who it is for. For developers who need accurate search over a large codebase straight from an agent with no cloud upload.
Good fit when
- You need to find a file by the query's intent rather than exact words
- You need go-to-definition or all uses of a symbol across a monorepo
- You want to cut tokens on irrelevant files during search
Not a fit when
- You need top raw speed on a very large repo rather than ranking
- A plain grep on exact string match is enough
Example request
Find where payment handling is implemented in the project and show the exact fileMCP
- Transport
- stdio
- Authentication
- not required
Limitations
On very large repos it is slower than grep and ripgrep on raw speed, since the bet is on result quality. Semantic reranking is enabled by a flag and pulls in local embeddings. The stated benchmarks and savings figures were measured by the author and are worth verifying on your own code.
How to disable. Remove the mantic block from your MCP client config. Uninstall the globally installed package via your package manager.
FAQ
Does data go to the cloud?
No, search runs fully locally with no data egress, including the optional local embeddings.
Is it only an MCP server?
No, there is a CLI and an MCP server, plus an editor extension; the server mode is started with the server argument.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server and CLI for code research: search and understand your project and all of GitHub with fewer tokens
An MCP server that indexes code with tree-sitter and returns individual functions and classes to the agent instead of whole files
A local codebase index for agents: hybrid search, dependency graphs and change impact analysis
The CodeAlive MCP server: the agent searches large codebases by meaning and exactly and follows call relations without a full read
codesearch searches code across several repositories with hybrid retrieval and runs fully offline, without a GPU or Docker.
An MCP server that turns a large codebase into a semantic feature graph for precise navigation and meaning-based search
README in short
The README describes Mantic as a structural code search engine for agents that puts relevance above raw speed. It infers intent from structure and metadata, drops irrelevant files before reading and thus saves tokens, and it also does go-to-definition and reference search across a monorepo. There is optional semantic reranking on local embeddings and code parsing via Tree-sitter. Everything runs locally with no data egress, installs as a CLI or connects as an MCP server over stdio via npx. MIT licensed.