Security

Secure by Design

Encryption standards, authentication, network isolation, and compliance certifications.

Secure by Design

Skytells implements defense-in-depth across every layer of the platform. Security is not an afterthought — it is a foundational design constraint applied from infrastructure provisioning through to API response delivery.


Encryption

Data in Transit

All communication with the Skytells API is encrypted using TLS 1.3. Connections using TLS 1.1 or earlier are rejected at the edge. This applies to:

  • API requests and responses (api.skytells.ai)
  • Webhook deliveries to your endpoints
  • CDN-delivered prediction outputs (delivery.skytells.cloud)
  • Dashboard and console sessions

Data at Rest

All persistent data — account information, billing records, and prediction metadata — is encrypted at rest using AES-256 encryption. Encryption keys are managed through hardware security modules (HSMs) with automatic key rotation.

Generated outputs (images, audio, video) stored during the 5-minute retention window are also encrypted at rest on the CDN origin.


Authentication

API Keys

Every API request must include a valid Bearer token in the Authorization header:

curl -H "Authorization: Bearer sk_live_your_api_key" \
  https://api.skytells.ai/v1/predictions

API keys follow these security practices:

PracticeDetail
Prefixed formatKeys use the sk_live_ prefix for production and sk_test_ for sandbox, making it easy to identify key scope.
Irreversible hashingOnly a hashed representation of your key is stored. The plaintext key is shown once at creation time.
Per-project scopingKeys can be scoped to specific projects, limiting blast radius if compromised.
Instant revocationCompromised keys can be revoked immediately from the dashboard. Revocation takes effect within seconds.

Rate Limiting

All API endpoints enforce rate limits to protect platform stability and prevent abuse. Rate limits are applied per API key and vary by account tier. See Rate Limits for detailed tier information.


Network Security

Infrastructure Isolation

Skytells inference workloads run in isolated compute environments with the following guarantees:

  • Tenant isolation — Each prediction executes in an ephemeral, single-tenant container. No shared memory, filesystem, or GPU context between tenants.
  • Network segmentation — Inference nodes are isolated in private subnets with no direct internet access. Egress is restricted to the CDN origin and internal services.
  • Edge protection — All public endpoints are behind a global edge network with DDoS mitigation, bot detection, and Web Application Firewall (WAF) rules.

Webhook Security

Webhook deliveries include an X-Skytells-Signature header for payload verification. Always validate this signature before processing webhook payloads. See Webhooks — Security for implementation details.


Compliance

Skytells maintains the following certifications and compliance standards:

StandardScopeDescription
SOC 2 Type IISecurity, Availability, ConfidentialityIndependent audit of security controls over a sustained period.
ISO 27001:2022Information Security ManagementCertified information security management system (ISMS).
ISO 27017Cloud SecurityControls specific to cloud service environments.
ISO 27018PII in Public CloudProtection of personally identifiable information in cloud infrastructure.

Data Processing

For enterprise customers, Skytells provides:

  • Data Processing Addendum (DPA) — GDPR-compliant processing terms available upon request.
  • Standard Contractual Clauses (SCCs) — For cross-border data transfers involving EU/EEA data subjects.
  • Subprocessor list — Maintained and updated with advance notice of changes.

Secure Development

PracticeDescription
Code reviewAll changes undergo mandatory peer review before deployment.
Dependency scanningAutomated vulnerability scanning of all third-party dependencies on every build.
Penetration testingRegular third-party penetration tests conducted against the API and infrastructure.
Incident responseDocumented incident response procedures with defined escalation paths and SLAs.
Responsible disclosureSecurity researchers can report vulnerabilities through skytells.ai/contact.

How is this guide?

On this page