Secure, production-ready OpenClaw BETA
Animus: the secure fork of OpenClaw for controlled, production-ready AI agents
Animus hardens the OpenClaw runtime with bounded execution, explicit approvals, and operator visibility — so agents work inside the limits you define.
Built on OpenClaw
We love OpenClaw. Animus is what it becomes for production.
Animus doesn't replace OpenClaw — it extends it. We kept everything that makes OpenClaw powerful and layered in the hardening, approvals, and memory architecture that production deployments need.
Powerful runtime and orchestration foundation. Flexible, battle-tested, the base we build on.
Risky capabilities are gated or removed. Agents operate within the limits you define.
People stay in control of what agents do — before it happens.
Clear logs for every action, decision, denial, and failure.
Memory becomes a real architecture layer — not just attached context.
Controlled, production-ready, self-hosted. OpenClaw, hardened for the teams that need it.
UMA: Unified Memory Architecture BETA
Animus implements UMA
UMA gives agents memory that survives a single conversation and retrieval that scales beyond a single prompt window.
Structured memory for agents that need to remember across conversations
Key Features:
- Stores facts, episodes, skills, and relationships in one structured memory layer
- Uses RLM (Recursive Language Model) to retrieve only what's relevant — not dump everything into the prompt
- Ranks memories by relevance, recency, and confidence so agents get better context
- Consolidates and prunes memory over time to stay accurate and manageable
- Separates agent, user, and project memory with clear ownership boundaries
Security changes we made
We rebuilt the runtime for secure, governed operation
Animus modifies the runtime, execution model, and operator workflows — not just the packaging.
Bounded execution
Animus keeps agent actions within clear limits instead of broad system access.
Approval-gated actions
Sensitive actions can be reviewed before execution, so people stay in control.
Audit-ready visibility
Clear logs make it easier to review actions, decisions, denials, and failures.
Documentation and repository
Explore the docs, review the architecture, and get the code
Review the runtime, security model, and architecture decisions behind Animus and UMA.