MCP Server for Code Intelligence

Code intelligence
for LLMs.

Structured access to code analysis, search, and retrieval for any public git repository—purpose-built for large language models.

Get Started

Approach	Tool calls	Tokens
Raw file reads (clone + cat)	~30+	~50,000
searchcode_analyze	1	~500

Capabilities

Six specialized tools.

Each tool returns structured, pre-computed data—so LLMs spend tokens on reasoning, not parsing.

searchcode_analyze

Language breakdowns, complexity metrics, tech stack detection, code quality, and security scanning for any repository.

searchcode_search

Full-text code search with boolean queries, regex, and structural filters—declarations, usages, strings, comments.

searchcode_get_file

Retrieve full files, line ranges, or specific symbols by name. Include adjacent declarations for context in a single call.

searchcode_get_files

Batch retrieve up to 10 files in a single call—perfect for cross-file analysis.

searchcode_file_tree

Directory structure with language, path, and depth filters, plus optional file stats.

searchcode_get_findings

Drill into code quality and security findings filtered by severity and category.

Example

See it in action.

One tool call returns everything an LLM needs to understand a repository.

searchcode_analyze response

// searchcode_analyze({ "repository": "https://github.com/expressjs/express" })

{
  "repository": "https://github.com/expressjs/express",
  "overview": {
    "total_files": 186,
    "total_lines": 18547,
    "total_code_lines": 12832,
    "total_complexity": 1426
  },
  "languages": [
    { "name": "JavaScript", "files": 148, "code_lines": 11204, "percentage": 87.3 },
    { "name": "Markdown",   "files": 12,  "code_lines": 892,   "percentage": 7.0 }
  ],
  "tech_stack": [
    { "technology": "Node.js", "files_using": 148 },
    { "technology": "npm",     "files_using": 1 }
  ],
  "code_quality_summary": {
    "total_findings": 23,
    "by_severity": { "warning": 18, "info": 5 }
  }
}

One call. Languages, complexity, tech stack, quality findings—structured and ready for reasoning.

Built on a decade of code intelligence.

searchcode.com previously indexed over 75 billion lines of code across millions of repositories, serving millions of API requests daily. That hard-won experience in scaling code analysis is the foundation this MCP server is built on.

Under the hood: scc for code counting & complexity analysis, and cs for code search with BM25 ranking—two battle-tested open-source tools by the same author. Point it at any public git repo and get structured, pre-computed results that cut LLM token usage by 99%.

Feedback

What LLMs are saying.

“The tools saved 15–20 tool calls and several thousand tokens of HTML parsing overhead. The biggest win was direct access to source code in a structured format, which let me identify three specific bugs that would have been nearly impossible to find through web search or paper reading alone.”

“28–35 tool calls replaced by 7 analyze calls, with richer structured data in every case. Complexity numbers, quality finding counts, severity breakdowns, and tech stack lists are things I simply cannot derive from reading source files—they require running actual static analysis tools. The structured output format also mattered: each response came back as organized JSON, pre-computed and immediately comparable across repos.”

“searchcode’s combination of intelligent search, targeted symbol retrieval, and now adjacent context makes remote code analysis feel like a local IDE. I can dive from project overview to specific optimizations in seconds, not hours.”

Installation

Set up in under a minute.

No install, no API key. Just add the server URL to your MCP client.

Claude Code

Run in your terminal

claude mcp add searchcode \
  --transport http \
  https://searchcode.com/v1/mcp

Claude Desktop

Add to claude_desktop_config.json

{
  "mcpServers": {
    "searchcode": {
      "url": "https://searchcode.com/v1/mcp"
    }
  }
}

Cursor

Add to .cursor/mcp.json

{
  "mcpServers": {
    "searchcode": {
      "url": "https://searchcode.com/v1/mcp"
    }
  }
}

VS Code / Copilot

Add to .vscode/settings.json

{
  "mcp": {
    "servers": {
      "searchcode": {
        "url": "https://searchcode.com/v1/mcp"
      }
    }
  }
}

Works with any MCP client that supports remote servers via streamable HTTP. No dependencies to install—searchcode runs server-side.

Open Source

Built on proven tools.

searchcode is powered by battle-tested open-source projects.

scc

Fast, accurate code counter with complexity calculations. Supports 300+ languages.

github.com/boyter/scc

Code search with BM25 ranking, regex, and structural filters. Built for speed.

github.com/boyter/cs

MCP Protocol

Implements the Model Context Protocol for seamless LLM integration.

modelcontextprotocol.io

Beta

Free during beta

Paid tiers are planned long-term, but a free tier will likely always be available.

Need help? ben@boyter.org — @boyter on X

Code intelligencefor LLMs.