Skip to content

Security Model

HyperMesh implements a layered security model designed for enterprise deployments in regulated industries (healthcare, defense, financial services).

HyperMesh uses API key-based authentication with role-based access control (RBAC).

RolePermissions
adminFull access: DDL, DML, key management, backup/restore
readwriteInsert, delete, compact, query
readonlyQuery and analytics only
Terminal window
# Create a new API key
hmdb keys create --role readwrite --description "ETL pipeline"
# List active keys
hmdb keys list
# Revoke a key
hmdb keys revoke hmdb_abc123

Keys are stored as bcrypt hashes in the schema database. The plaintext is shown only at creation time.

  • Database files (TPI, FMI, PSI, WAL) are stored as binary indexes on disk
  • For encryption at rest, deploy on an encrypted filesystem (LUKS, AWS EBS encryption, Azure Disk Encryption)
  • Schema metadata (SQLite) supports SQLCipher for column-level encryption
  • The REST API server (hmdb serve) supports TLS via --tls-cert and --tls-key flags
  • Client SDKs verify TLS certificates by default (verify_ssl=True)
Terminal window
hmdb serve --tls-cert /etc/ssl/certs/hypermesh.pem \
--tls-key /etc/ssl/private/hypermesh.key \
--port 8443
  • Bind to specific interfaces with --host (default: 127.0.0.1)
  • Rate limiting configurable via HMDB_RATE_LIMIT environment variable
  • Health endpoints (/health/live, /health/ready) do not require authentication

When HMDB_AUDIT_LOG is set, all mutating operations are logged with:

  • Timestamp (UTC)
  • API key identifier (not the key itself)
  • Operation type (INSERT, DELETE, CREATE TABLE, etc.)
  • Affected table and record count

HyperMesh is designed to support compliance with:

  • SOC 2 Type II — audit logging, access controls, encryption
  • HIPAA — PHI isolation via table-level access, encryption at rest/in transit
  • FedRAMP — deployable in GovCloud regions with FIPS 140-2 validated TLS

For detailed compliance documentation and attestation letters, contact security@hypermeshai.com.

Report security vulnerabilities to security@hypermeshai.com. We follow a 90-day disclosure timeline and acknowledge reports within 48 hours.