ai-image-analysis-mcp: Instructions file for Gemini CLI

I have reviewed the codebase and have a good understanding of the project. Here's a summary of my findings:

Project Overview

This project, "AI Image Analysis MCP v2.0," is a robust and feature-rich application for analyzing images using Google's Gemini 2.0 Flash model. It's designed with a security-first approach and offers multiple deployment and access methods.

Key Architectural Points

• Modular Design: The src directory is well-organized into modules, separating concerns like Gemini integration (gemini-analysis.ts), security (security.ts), Supabase interactions (supabase-upload.ts), and data types (types.ts). This makes the code maintainable and easy to understand.
• Multiple Entry Points: The application can be run in several ways:
  • As a local MCP server (src/index.ts).
  • As a proxy to a Supabase Edge Function (src/mcp-supabase-proxy.ts).
  • Directly via an HTTP client (src/supabase-mcp-client.ts and src/api-client.ts).
• Refactoring: The presence of index-original.ts suggests a significant refactoring effort to achieve the current modular structure. The new index.ts is much cleaner and delegates logic to the appropriate modules.
• Security Focus: Security is a core tenet of this application. The security.ts module includes functions for:
  • Input sanitization.
  • Prompt injection detection.
  • PII (Personally Identifiable Information) detection.
  • Rate limiting.
  • Auditing requests.
• Image Integrity: The integrity.ts and utils/mime-detection.ts files show a focus on correctly handling image files, preserving their formats, and verifying their integrity using checksums. This is a crucial feature for a reliable image processing pipeline.
• URL Fetching: The utils/url-fetcher.ts module provides a secure way to fetch images from URLs, with protections against common vulnerabilities like SSRF (Server-Side Request Forgery).

Code Highlights