Awesome MCP › Software Development

marcoaapfortes/Mantic.sh

⭐ 556 TypeScript added to this list on 2026-01-11 repository created 2026-01-04

Mantic.sh is a context-aware structural code search engine designed specifically for AI agents, prioritizing relevance and intent understanding over raw search speed. It operates by inferring intent from file structure and metadata rather than brute-force content scanning, enabling faster and more meaningful code retrieval. The project supports advanced search features such as CamelCase detection, exact filename matching, path sequence matching, and semantic multi-word queries, which significantly improve the accuracy and relevance of search results compared to traditional tools like grep, ripgrep, and fzf. Mantic.sh is optimized for AI agents by reducing unnecessary context retrieval overhead, which results in retrieval speeds typically under 500 milliseconds for most repositories and under 4 seconds for very large monorepos like Chromium. It also reduces token usage by filtering out irrelevant files before reading, enhancing efficiency and privacy since it runs entirely locally with zero data egress. The tool supports session management with context carryover, boosting results based on previously viewed files, and includes a learning system that caches successful search patterns for adaptive ranking. The project is production-ready and has been tested on large codebases, demonstrating superior result quality and 100% multi-repo accuracy. It offers a zero-query mode to provide insights such as modified files and related dependencies without explicit queries. Mantic.sh is fully compatible with the Model Context Protocol (MCP), enabling integration with MCP-compatible tools like Claude Desktop, Cursor, and VS Code. While Mantic.sh is slower than some traditional search tools in raw speed, it excels in delivering highly relevant and contextually accurate results, making it ideal for AI agents and developers who need to understand code structure and intent. It is less suited for quick text searches or extremely large repositories where speed is critical. The project is open-source and provides easy installation options via npm or from source, with detailed documentation and benchmarks available.

https://github.com/marcoaapfortes/Mantic.sh

adaptive-rankingai-agentai-agentscamelcase-detectionclaudecode-retrievalcode-searchcode-structure-understandingcodebase-searchcontext-aware-code-searchcontext-retrievalcursordeveloper-toolsexact-filename-matchingimpact-analysislocal-firstmcpmcp-servermodel-context-protocolmulti-repo-accuracyopen-sourcepath-sequence-matchingprivacyragrelevance-rankingsemantic-searchsession-managementstructural-code-searchtypescriptzero-query-mode