sourcerer-mcp

🥈Silver

Sourcerer MCP enables AI agents to perform semantic code search and navigation, reducing token usage by jumping directly to relevant functions, classes, and code chunks. Benefits developers and operations teams by improving efficiency in code analysis and navigation tasks. Connects to Claude agents via Model Context Protocol.

111100Updated 2mo ago

Intermediate30min to implementautomation

Saves ~90 min per use

Quick InstallView Source

git clone https://github.com/st3v3nmw/sourcerer-mcp.git

Works with:

Claude

Overview

About This Skill

How to Use

[{"step":"Install and configure sourcerer-mcp in your development environment. Ensure it has access to your project's codebase by setting the correct [PROJECT_ROOT] path in the MCP server configuration.","tip":"Use `mcp-server-sourcerer --init` to generate a default configuration file, then update the `source_directories` array to include your project's root and any relevant submodules."},{"step":"Define your search parameters by specifying the [FUNCTION_NAME] or [CLASS_NAME] you want to analyze. For broader searches, use [REGEX_PATTERN] to match multiple related functions.","tip":"Combine multiple placeholders for complex queries, e.g., `[FUNCTION_NAME]_test` to find test cases for a specific function. Use sourcerer-mcp's `--context-lines` flag to adjust the amount of surrounding code returned."},{"step":"Execute the search using your preferred interface (Claude, terminal, or IDE plugin). For large codebases, consider narrowing the search scope to specific directories or file types.","tip":"Use `--include` and `--exclude` flags to filter results by file extensions (e.g., `--include '*.py'` or `--exclude '**/tests/**'). For multi-project analysis, chain searches using the `--project` flag."},{"step":"Review the structured output and categorize the results by [CONTEXT_TYPE] or other relevant criteria. Use the extracted snippets to inform refactoring, debugging, or documentation tasks.","tip":"Export the results to a file using `--output json` for further processing. Use the `file_path` and `line_number` fields to quickly navigate to specific code locations in your IDE."},{"step":"Iterate by refining your search parameters based on the initial results. For example, search for related functions or classes mentioned in the context snippets to build a comprehensive understanding of the codebase.","tip":"Use sourcerer-mcp's `--related` flag to automatically find functions called by or calling your target function, helping you map dependencies and usage patterns."}]

Setup & Installation

Quick Install

No install command available. Check the GitHub repository for manual installation instructions.

Alternative Install (Git Clone)

git clone https://github.com/st3v3nmw/sourcerer-mcp

Requirements

Claude Code or compatible AI agent
Works with: Claude

Quick Start Guide

Install the Skill

Copy the install command above and run it in your terminal.

Open Your AI Agent

Launch Claude Code, Cursor, or your preferred AI coding agent.

Try It Out

Use the prompt template or examples below to test the skill.

Customize

Adapt the skill to your specific use case and workflow.

Usage Examples

Prompt Template

Use sourcerer-mcp to analyze the [PROJECT_NAME] codebase and identify all instances where [FUNCTION_NAME] is called or referenced. For each instance, extract the surrounding 5 lines of context and categorize them by [CONTEXT_TYPE: e.g., 'error handling', 'performance optimization', 'legacy code']. Return the results as a structured JSON report with file paths, line numbers, and categorized snippets.

Example Output

```json
{
  "search_results": {
    "project_name": "E-Commerce Platform",
    "function_name": "calculate_discount",
    "total_instances": 12,
    "by_context_type": {
      "error_handling": [
        {
          "file_path": "/src/services/discount_service.py",
          "line_number": 45,
          "context": [
            "try:",
            "    discount = calculate_discount(user, cart)",
            "    if discount < 0:",
            "        raise ValueError('Invalid discount calculated')",
            "except ValueError as e:"
          ]
        }
      ],
      "performance_optimization": [
        {
          "file_path": "/src/utils/pricing_utils.py",
          "line_number": 89,
          "context": [
            "def apply_bulk_discounts(items):",
            "    discounts = [calculate_discount(item) for item in items]",
            "    return sum(discounts)",
            "    # Note: Parallel processing could improve performance here"
          ]
        }
      ]
    }
  }
}
```

**Summary:** The `calculate_discount` function is referenced 12 times across the codebase, with 3 instances in error handling contexts and 2 in performance optimization contexts. The most complex usage is in `/src/services/discount_service.py` where it interacts with user data validation. The performance-critical path in `/src/utils/pricing_utils.py` suggests potential for parallel processing optimization.