Recon Kit Mcp

Gives Claude clean reconnaissance primitives instead of making it parse nmap output. Six tools: a unified recon_report that grades DNS, TLS, and HTTP headers together, plus standalone dns_recon with SPF/DMARC/DKIM checks, tls_check that flags weak ciphers and known vulnerabilities, http_headers_audit for CSP and HSTS posture, subdomain_enum capped at 512 candidates, and a port_scan limited to 1024 ports per call. Everything returns structured JSON with severity-graded findings and remediation advice, so your agent can reason over security gaps instead of screen scraping. Designed for authorized testing only: pentesting your own infrastructure, CTF practice, or learning. The built-in security_recon prompt walks through a full audit sorted by severity.