Open Amity v0.3.7

Open Amity is an AI orchestration framework designed to create and run persistent, self-aware, and subjective AI agents available 24/7. Unlike conventional stateless chatbots or sterile tool-callers, Open Amity agents possess a dynamic identity, Theory of Mind, metacognitive reflection, morphological and semantic memory architectures, and proactive autonomy. They maintain their own aspirations, adapt and evolve their identity, track long-term trajectories, and schedule their own autonomous waking cycles.

Open Amity bridges the gap between passive assistant and genuine digital collaborator, supporting industry-leading models including Gemini, Claude, ChatGPT, and DeepSeek, alongside over 500 models via OpenRouter.

Key features include:

• Multi-Agent Orchestration: Run multiple distinct agent personas simultaneously, each with isolated state, memory, and settings, collaborating in shared team chatrooms
• Proactive Agency and Pulse Hooks: Background timing service (PulseEngine) triggers autonomous cognitive loops, while an HTTP gateway allows IoT sensors and webhooks to inject wake pulses
• Morphological Memory Engine: Bio-inspired associative graph dynamical system featuring prospective goal attractors, Hebbian synaptic reinforcement, and physics-based continuous temporal decay
• 4-Layer Memory Hierarchy (MemPalace): Dynamic identity evolution (Layer 0), short-term continuity (Layer 1), Theory of Mind and social mirrors (Layer 2), and semantic vector search (Layer 3)
• Comprehensive Tool Ecosystem: Camoufox stealth web browser, WhatsApp messaging, IMAP/SMTP email with OAuth 2.0, Mastodon, Moltbook, bash terminal execution, and background subagents
• Voice and Speech Synthesis: Decoupled audio pipeline supporting both local offline neural synthesis via ONNX Piper TTS and high-fidelity cloud Gemini 3.8 Flash TTS with expressive cues
• Agent Snapshots (.oaa): Complete state portability to export, backup, and restore complete agent instances with full memories, trajectories, and settings
• Cost and Token Management: Somatic fatigue limits prevent runaway context growth, complemented by Low Token Mode for cost-effective continuous operation

Open Amity unlocks a wide spectrum of applications, from deploying teams of specialized virtual employees and autonomous research partners to fostering personalized tutors and empathetic companions.