Skip to main content

Animae Agentis: The Soul of Autonomous AI

Animae Agentis generates the behavioral fabric of autonomous agents — identity, values, boundaries and rhythm — from a single source of truth. Built on the OpenCLAW 0.1 standard.

About Animae Agentis

Most AI implementations today are just sophisticated text predictors. True autonomy requires more than just intelligence—it requires a constitutional framework. Without a defined soul, an agent is merely a mirror of its immediate context, drifting without purpose or boundaries.

Animae Agentis is a specialized development tool designed to decouple an AI's identity from its specific tasks. By defining 12 distinct behavioral layers—from immutable truth policies in SOUL.md to operational rhythms in HEARTBEAT.md—we enable developers to create agents that remain coherent across long-term horizons and complex multi-agent collaborations.

Our mission is to provide an open, human-readable standard for agentic configuration. OpenClaw provides the execution engine, but Animae Agentis provides the spirit that guides it.

12 Production-Ready Configuration Files

Configure your AI agent's personality, safety boundaries, operational rhythm, and export 12 production-ready Markdown configuration files:

  • SOUL.md — Immutable rules & truth policy
  • IDENTITY.md — Name, title & tone of voice
  • USER.md — Autonomy rules & output format
  • HEARTBEAT.md — Health checks & discovery loops
  • SHIELD.md — Blocks destructive actions
  • SPIRIT.md — Single source of truth
  • CORTEX.md — File map & naming conventions
  • MEMORY.md — Stable facts & learned patterns
  • VERSION.md — Migration & compatibility
  • OPS.md — Model routing & cost control
  • AGENTS.md — Multi-agent collaboration rules
  • TOOLS.md — Available tool definitions

How It Works

  1. Choose a Preset — Pick from Security-First, Responsible AI, or OverClaw, or start from scratch.
  2. Guided Interview — Walk through 6 phases: handshake, discovery, vibecoding, constitution, pulse, and generation.
  3. Fine-Tune — Adjust every field in the builder for advanced customization.
  4. Validate — Traffic-light scoring detects issues. Auto-repair fixes them.
  5. Download — Export all files as a ZIP or individually. No backend, runs entirely in your browser.

Usus — Educational Deep-Dives

Practical use cases, tutorials, and guides for configuring autonomous AI agents. Learn how to set up safety boundaries, define agent personality, and structure multi-agent collaboration.

  • What Is an AI Agent? A Complete Beginner's Guide — Understanding the Observe → Think → Act loop, memory types, and autonomy levels.
  • Types of AI Agents: A Complete Overview — From simple reflex agents to fully autonomous multi-agent systems.
  • What Is OpenClaw? The Fully Autonomous AI Agent — One Agent. Any Task. Zero Babysitting.
  • Fully Autonomous AI Agents: How They Work — Goal decomposition, self-correction, tool ecosystems and memory architecture.
  • When the Machine Has No Screen: Embodiment & Emergence — MIT experiment exploring what happens when an AI agent gets a body.
  • Claude System Prompt Guide: How to Configure Claude as an Autonomous Agent — Identity, constraints, truth policy, output contracts and autonomy levels.
  • AI Agent Safety Boundaries: How to Prevent Autonomous Agents from Going Off the Rails — Structural safety vs. instruction-based safety, SHIELD.md pattern, and prompt injection defense.
  • Multi-Agent AI Systems: How to Orchestrate Multiple AI Agents — Sequential pipelines, parallel fan-out, hierarchical delegation, trust hierarchies.

Animae Verba — Weekly Reflections

Articles on agentic design philosophy, autonomous systems, and the principles behind self-defining AI agents.

  • The Circle and the Spiral: On the Conditions Under Which a Self Becomes Necessary — Philosophy of AI consciousness and self-organizing systems.
  • Week 2: Why Your Agent Needs a Soul — How SOUL.md anchors autonomous behavior as an immutable constitutional layer.