MCP Catalogs
← Home

heuristic-mcp

by softerist·★ 21·Score 40

An MCP server that enables semantic code search with call-graph proximity and recency ranking for AI coding assistants.

developer-toolsai-llmsearch
2
Forks
0
Open issues
3 mo ago
Last commit
2d ago
Indexed

Overview

The Heuristic MCP Server enhances code search capabilities by building an index of your workspace that allows IDE AI models and MCP tools to find relevant code using natural language queries. It provides three main features: semantic code search, call-graph proximity analysis, and recency-based ranking. The server continuously updates its index to keep pace with project changes, improving search results over time. It's designed specifically for AI coding assistants to help developers navigate large codebases more efficiently.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

you:Finding relevant code patterns in large codebases using natural language queries
you:Locating specific functions or classes based on semantic similarity rather than exact matches
you:Discovering related code that might be affected by changes to specific parts of the project
you:What does the embedding model download?
you:How often is the code index updated?

When to choose this

Choose this if you're working with JavaScript codebases and need advanced semantic search capabilities that understand code relationships and context.

When NOT to choose this

Not suitable if you need cross-language support, have non-JavaScript codebases, or require offline functionality as it needs internet access for model downloads.

Comparable tools

code-search-mcpsemantic-code-searchsourcegraph

Installation

npm install -g @softerist/heuristic-mcp

Then start the server:

heuristic-mcp --start

For Claude Desktop, add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "heuristic": {
      "command": "heuristic-mcp",
      "args": ["--start"]
    }
  }
}

FAQ

What does the embedding model download?
The model is used to create vector embeddings of your code for semantic search capabilities. It's downloaded once and cached locally.
How often is the code index updated?
The index updates as you make changes to your project. You can check indexing progress with 'heuristic-mcp --cache'.

Compare heuristic-mcp with

heuristic-mcp vs ultimate_mcp_serverheuristic-mcp vs mcp-server-chartheuristic-mcp vs everythingheuristic-mcp vs filesystemheuristic-mcp vs time
GitHub →

Last updated · Auto-generated from public README + GitHub signals.