v0.11.14

What's New

Simulation Engine

• 20-probe behavioral simulation battery across 7 attack categories
• Mock tool environment (API, database, memory, credential, communication)
• 3-layer architecture: NanoMind TME (8ms) -> targeted probes (3s) -> full simulation (30s)
• hackmyagent secure --deep runs full behavioral simulation

Adaptive Attack Engine

• hackmyagent red-team <file> - NanoMind-powered adaptive red team
• Target-specific payload generation from semantic vulnerability surface
• Iterative attack with defense mapping (up to 5 iterations per category)
• Auto-exports training data from every attack session

NanoMind Default-On

• NanoMind semantic analysis auto-enabled when daemon is available
• No flags needed - smart defaults based on what's available
• --static-only for CI/deterministic mode

CLI Improvements

• Version now reads from package.json (no more stale version strings)
• hackmyagent explain <findingId> with static fallback when daemon unavailable
• Simulation skips gracefully when no skill artifacts found

Bug Fixes

• Fixed duplicate attack command registration (renamed adaptive engine to red-team)
• Fixed --deep hanging when no LLM backend available
• Fixed ASFF output using hardcoded version string