ATP Runtime Guard for OpenClaw
Drop ATP's runtime guard into your OpenClaw workflows to add quantum-safe cryptography, tool-level security, and dynamic trust scoring.
Architecture
Layered Security Approach
ATP wraps every layer of your OpenClaw system with quantum-safe security
┌─────────────────────────────────────────────────────────┐
│ OpenClaw Agent Crew │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ │
│ │ Agent A │──│ Agent B │──│ Agent C │ │
│ │ (DID:..1)│ │ (DID:..2)│ │ (DID:..3)│ │
│ └─────┬────┘ └─────┬────┘ └─────┬────┘ │
└────────┼─────────────┼─────────────┼───────────────────┘
│ │ │
▼ ▼ ▼
┌─────────────────────────────────────────────────────────┐
│ ATP OpenClaw Security Layer │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ │
│ │ Agent │ │ Tool │ │ Graph │ │
│ │ Wrapper │ │ Wrapper │ │ Validator│ │
│ └──────────┘ └──────────┘ └──────────┘ │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ │
│ │ Policy │ │ Monitor │ │ Secrets │ │
│ │ Engine │ │ +Lunary │ │ Manager │ │
│ └──────────┘ └──────────┘ └──────────┘ │
└────────────────────┬────────────────────────────────────┘
│
▼
┌───────────────────────┐
│ ATP Core SDK │
│ - Identity Service │
│ - Trust Scoring │
│ - Audit Logging │
└───────────────────────┘Key Features
Quantum-Safe Identities
Every agent gets a cryptographic DID with hybrid Ed25519 + Dilithium3 signatures
- Post-quantum secure by default
- Decentralized identity (DID)
- Automatic key rotation
Tool-Level Security
ATP intercepts every tool call with 5-step validation
- Authentication checks
- Trust level validation
- Permission verification
- Rate limiting & DLP
Graph Validation
Policy-based constraints on agent interactions before execution
- Cycle detection
- Chain depth limits
- Data flow policies
Dynamic Trust Scoring
Agent trust levels adjust based on behavior and success rate
- 0.0-1.0 trust scores
- Automatic adjustments
- Trust-based access control
Observability
Integration with Lunary for metrics and anomaly detection
- Real-time monitoring
- Anomaly detection
- Automatic trust updates
Secret Management
Short-lived, scoped credentials for external services
- Time-limited tokens
- Scope-based access
- Automatic rotation
Quick Start
Install Package
Add ATP Runtime Guard to your OpenClaw project
npm install @atp/openclaw-atp atp-sdk # or with yarn yarn add @atp/openclaw-atp atp-sdk # or with pnpm pnpm add @atp/openclaw-atp atp-sdk
Pre-Configured Security Profiles
strictDev
Development and testing environment
- Min trust:0.6
- Allow cycles:Yes
- Max depth:10
productionFinance
High-security financial applications
- Min trust:0.95
- Require MFA:Yes
- Allow cycles:No
- Max depth:5
piiWorkflow
GDPR/CCPA compliant workflows
- Min trust:0.9
- Require consent:Yes
- Data minimization:Yes
- Allow sharing:No
researchWorkflow
Balanced security for research and exploration
- Min trust:0.7
- Allow cycles:Yes
- Max depth:8
Use Cases
Financial Trading
Secure multi-agent trading systems
Customer Service
Multi-tiered support automation
Why ATP Runtime Guard?
Cryptographic Agent Identity
ATP Runtime Guard gives every OpenClaw agent a cryptographic DID, so every action is attributable, verifiable, and auditable — not just assumed.
Quantum-Safe by Default
ATP wraps OpenClaw agent communications with ML-DSA (Dilithium) signatures, future-proofing your workflows against quantum threats without code changes.
Trust Without Blind Faith
ATP's dynamic trust scoring lets your OpenClaw workflows adapt in real time — restricting low-trust agents, elevating proven ones, automatically.
OpenClaw handles the "how" of multi-agent work. ATP Runtime Guard handles the "who" and "should they".