Skip to main content
Kwala follows a zero-trust, verifiable execution approach that builds security guarantees across its infrastructure. Every layer of the system implements specific security measures to protect user assets and ensure trustless execution.

Security by layer

Kwala implements security measures at every architectural layer:
Security layerMeasures
User layerSelf-custody of keys, signed intents, non-repudiation
Node layerPermissioned entry, KMS-backed keys, sandboxed execution, mTLS
Workflow layerImmutable YAML intents, conditionally executable logic, execution logs
Verifier layerWatchdog verification, proof-based dispute, slashing mechanism
Infrastructure layerDistributed infra, autoscaling, rate-limiting, failover and redundancy

Core security measures

Sandboxed execution

Workflows run in isolated environments that prevent unauthorized access to system resources or external entities.
  • Each workflow executes in its own container
  • No access to host filesystem or network outside defined actions
  • Resource limits prevent denial-of-service attacks
  • Execution environments are destroyed after completion

Deterministic workflows

Workflows execute identically across all nodes, producing consistent, verifiable results regardless of the node processing them.
  • Same inputs always produce same outputs
  • No reliance on node-specific state or randomness
  • Enables independent verification by any party
  • Ensures fair and predictable execution

Cryptographically signed actions

Every action is signed and logged, creating an immutable audit trail traceable to the original user intent.
┌─────────────────┐     ┌─────────────────┐     ┌─────────────────┐
│   User Intent   │────▶│   Node Action   │────▶│   On-Chain Log  │
│   (signed)      │     │   (signed)      │     │   (immutable)   │
└─────────────────┘     └─────────────────┘     └─────────────────┘
        │                       │                       │
        └───────────────────────┴───────────────────────┘
                    Cryptographic chain of custody

Immutable workflow history

All executions are permanently recorded on the Kalp Chain to enable audits and verification. You can use intent hashes, execution logs, and reports to verify historical workflow executions.
Record typePurpose
Intent hashesProve user authorization for specific workflows
Execution logsDetail every action taken during execution
Verifier reportsAttest to correctness of execution

Independent verifier nodes

Lightweight validators recompute workflow logic and verify the correctness of executor actions and outputs.
  • Operate independently from execution nodes
  • Randomly assigned to prevent collusion
  • File disputes when deviations detected
  • Participate in slashing decisions

Permissioned nodes with KMS-backed keys

All nodes are vetted through Kalp Governance, with private keys stored in non-extractable Key Management Systems that perform signing operations in hardware enclaves.
  • Node operators cannot access raw private keys
  • Signing happens inside hardware security modules
  • Keys are bound to specific node identities
  • Audit trails for all signing operations

Encrypted inter-node communication

All node-to-node communication uses mutual TLS (mTLS) with node-specific certificates for authentication and encryption.
  • TLS 1.3 encryption for all traffic
  • Mutual authentication prevents impersonation
  • Certificate-based node identity verification
  • Protection against man-in-the-middle attacks

User layer security

Users maintain complete control over their security:

Self-custody

Users hold their own private keys—Kwala never has access

Signed intents

Every workflow requires explicit user signature

Non-repudiation

Cryptographic proof of user authorization

Revocation

Users can deactivate workflows at any time

Infrastructure security

The Kwala infrastructure implements enterprise-grade security controls:
ControlImplementation
Distributed infrastructureNodes spread across multiple regions and providers
AutoscalingDynamic capacity to handle load spikes
Rate limitingProtection against abuse and DoS attacks
Failover and redundancyAutomatic recovery from node failures
Network isolationSegmented networks for different functions
Monitoring and alertingReal-time detection of anomalies

Next steps