
Smart Coding MCP
par omar-haris
Semantic code search for AI assistants using local embeddings — find code by meaning, not just keywords.
What it does
Smart Coding MCP connects AI coding assistants to your codebase with intelligent semantic search. Instead of matching exact keywords, it indexes your project with AI embeddings so you can ask natural-language questions like "where do we handle authentication?" and find relevant code even when the terminology differs. Everything runs locally — no API calls, no data leaving your machine.
Tools
- a_semantic_search — Find code by meaning using AI embeddings with hybrid semantic + exact-match ranking
- d_check_last_version — Look up the latest version of any package across 20+ registries (npm, PyPI, Crates.io, Maven, Go, RubyGems, and more)
- b_index_codebase — Trigger a full reindex of your project; supports progressive indexing so search works immediately
- c_clear_cache — Delete the embeddings cache and force a fresh reindex
- e_set_workspace — Switch workspace paths at runtime for multi-project or monorepo workflows
- f_get_status — Check server health, indexing progress, model config, and cache stats
Installation
Add to your claude_desktop_config.json:
{
"mcpServers": {
"smart-coding-mcp": {
"command": "smart-coding-mcp",
"args": ["--workspace", "/absolute/path/to/your/project"]
}
}
}
For multi-project setups, add separate entries per workspace. Supports VS Code Copilot, Cursor, Windsurf, OpenCode, and Claude Desktop.
Supported hosts
Claude Desktop, VS Code Copilot, Cursor, Windsurf, OpenCode, Raycast
Installation rapide
npx smart-coding-mcp --workspace /path/to/projectInformations
- Tarification
- free
- Publié






