Sift 2.2.0

Adds the Scan tab: passive analysis of captured traffic and an active vulnerability scanner (reflected XSS, SQL injection, SSTI, command injection, path traversal, open redirect, SSRF, CRLF, XXE, NoSQL injection) with baseline comparison, content discovery and parameter mining.

Adds an AI layer (OpenAI-compatible, Anthropic and Gemini, including local models via Ollama/vLLM) for triage, correlation, test recommendations and verification, with a model-per-task routing and fallbacks.

Adds a token-gated MCP server so an external AI agent can read findings and traffic and drive scans.

Findings carry severity, confidence, evidence, reproduction steps and remediation, and link back to the exact request.