mcp-recon

mcp-recon is a lightweight yet powerful conversational reconnaissance interface built on the Model Context Protocol (MCP). By exposing ProjectDiscovery’s and tooling through a simple text‑based API, it lets AI assistants—such as Claude Desktop—perform deep web domain analysis without leaving the chat. The server bridges natural language commands with HTTP infrastructure discovery, making it trivial for developers to incorporate live reconnaissance into automated workflows.

The core problem mcp-recon solves is the friction between AI conversation and actionable network intelligence. Traditional reconnaissance tools require command‑line knowledge, manual parsing of output, or custom scripts to integrate with an assistant. mcp-recon eliminates these hurdles by providing ready‑made MCP prompts (e.g., , ) that translate user intent into precise tool invocations. Developers can now ask an assistant to “scan example.com for status codes” or “generate a Katana crawl command,” and the server handles execution, parsing, and formatting of results.

Key capabilities include two reconnaissance modes—lightweight and full. The lightweight scan offers quick HTTP fingerprinting, returning status codes, server headers, and basic technology detection. The full scan dives deeper, collecting page previews, TLS certificates, detailed headers, and extensive tech stacks. Additionally, the server offers a standalone ASN lookup service that resolves IPs, ASNs, or organization names using . These features are packaged as distinct MCP prompts, enabling modular use and chaining within larger AI-driven pipelines.

Real‑world scenarios that benefit from mcp-recon are plentiful. Security teams can embed the server into incident‑response chatbots to rapidly verify host health or gather certificate information during threat hunting. DevOps workflows may use the ASN lookup to validate third‑party services or detect unexpected routing changes. Penetration testers can generate Katana crawl commands on the fly, streamlining target discovery. Because mcp-recon runs in a Docker container, it can be deployed securely on isolated networks or cloud environments, ensuring that reconnaissance data never leaks beyond the intended scope.

Integrating mcp-recon into an AI workflow is straightforward: add the server to your MCP configuration, reference its prompts in assistant instructions, and let the model orchestrate calls. The server’s automatic handling of ’s stdin/stdout quirks and its pre‑built prompt library reduce boilerplate, allowing developers to focus on higher‑level logic rather than tool orchestration. In short, mcp-recon turns complex HTTP reconnaissance into a conversational, repeatable service that enhances productivity and reduces operational friction for AI‑powered security and development tasks.