In-Memoria: Instructions file for GitHub Copilot

GitHub Copilot Instructions for In Memoria

This repository includes In Memoria, an intelligent MCP (Model Context Protocol) server that provides codebase intelligence through semantic analysis, pattern recognition, and smart navigation.

⚠️ CRITICAL: Path Parameter Usage

ALWAYS provide absolute paths to In-Memoria tools. NEVER rely on default values.

Why This Matters

In MCP server context, process.cwd() is unpredictable and may point to the wrong directory (like /home/user instead of your project). Always specify the path explicitly to avoid analyzing the wrong codebase or creating databases in unexpected locations.

// ❌ WRONG - May use incorrect directory
await use_mcp_tool('in-memoria', 'get_project_blueprint', {
  includeFeatureMap: true
});

// ✅ CORRECT - Always specify path explicitly
await use_mcp_tool('in-memoria', 'get_project_blueprint', {
  path: '/absolute/path/to/project',  // Use workspace root
  includeFeatureMap: true
});

Getting the Project Path

• In VS Code: Use ${workspaceFolder} or workspace root API
• Ensure it's an absolute path (starts with / on Unix, C:\ on Windows)
• Be consistent across all tool calls in a session
• Verify the path exists before calling tools

Path Convention for All Tools

Every tool that accepts a path parameter should receive:

• Absolute paths to the project root directory
• Same path throughout the entire session
• No relative paths like . or ./src (resolve them first)

---

Core Capabilities

In Memoria learns from codebases and provides:

• Instant project context - Tech stack, entry points, key directories, architecture overview
• Semantic search - Find code by meaning, not just keywords
• Pattern recognition - Discover coding patterns and best practices
• Smart file routing - Navigate to relevant files from vague requests
• Coding approach predictions - Get implementation suggestions based on learned patterns