TITAN v7.0.0 "Independence" 🎆

Local-first is independence: your AI, your hardware, your data.

Headliners

💪 Muscle Memory — the first trustworthy automatic self-improvement in any agent framework. TITAN notices your repeated workflows, teaches itself a parameterized skill, and proves it works by replaying it against your real usage through a deterministic eval harness before you ever see it. One click to adopt → instant slash command. Nothing auto-adopts, side-effectful workflows are never mined, dismissals are remembered.
👋 Welcome Mode — first run in about a minute. The gateway always boots; the dashboard walks you into connecting a model (one click if Ollama is running). No Docker, no YAML, no terminal ceremony.
🫀 Alive by default — heartbeat notes the mascot speaks, proactive memory, Soma proposals, real reminders — plus a deterministic honesty guard: TITAN can never claim it scheduled something it didn't.
🪵 The Living Desk — home is a canvas, the mascot is a creature that wanders, celebrates, and delivers. Five doors, not 48.
🧠 Model-agnostic harness — per-model capability registry, 36 providers, adaptive token ceilings. Benchmarked: best local = qwen3-coder-next-class, best cloud = GLM-5.1 / Kimi K2.6 (see benchmarks/MODEL_COMPARISON.md).
♿ WCAG 2.1 AA with a contrast-audit harness proving every theme.

Full details in CHANGELOG.md. Verified: 8,122 tests / 0 failing; 26-agent adversarial review on Muscle Memory; live-proven on real hardware.

Quick start: npm install -g titan-agent && titan gateway — the desk walks you in. 🎆