Infrastructure
MNKI AgentOS runs on Cloudflare's global edge network using Workers for compute, D1 for our database, and KV/R2 for storage. We build on infrastructure providers (Cloudflare, Amazon Web Services, Stripe) that maintain independent security certifications and operate hardened, regularly-audited data centers.
Encryption
- •All traffic is encrypted in transit with TLS.
- •Data at rest is encrypted by our infrastructure providers.
- •Credentials and tokens for Connected Services (e.g. Google, Slack) are encrypted at the application layer with authenticated AES-GCM encryption before storage, and decrypted only at execution time to perform the task you authorized.
Authentication & sessions
- •We use passwordless authentication with one-time codes delivered by email — there is no password for attackers to phish or reuse.
- •Sessions are managed with secure, httpOnly cookies and server-side validation, with revocation on logout.
- •Administrative access is restricted to an explicit allowlist and gated on every privileged request.
Tenant isolation
Every Customer belongs to an organization, and data access is scoped to that organization on every query. Agent deployments, execution logs, configurations, and per-deployment memory are isolated by organization and deployment, so one tenant cannot access another's data.
Payments
Payments and payouts are handled by Stripe, a PCI-DSS Level 1 certified provider. We never see or store full card numbers; card data is sent directly to Stripe. Builder payouts use Stripe Connect with Stripe-managed identity verification.
AI & data handling
- •Agent inputs are sent to model providers and connected tools only as needed to perform the task you configured.
- •We use providers that contractually limit use of submitted data, and we do not use your private agent inputs or outputs to train our own models without consent.
- •Connected Service access uses scoped authorization and can be revoked by you at any time.
Builder agent review
Agents submitted by third-party Builders go through a review process before going live, covering quality, safety, and policy. We may suspend or remove Agents that violate our policies or behave unexpectedly. Builders are contractually required to handle Customer data and credentials securely and only as needed to perform their function.
Operational practices
- •Least-privilege access to production systems and secrets, managed as encrypted secrets rather than in source code.
- •Separate staging and production environments with environment-scoped configuration.
- •Logging and monitoring of agent runs, with dead-letter handling and alerting for failures.
- •Version-controlled infrastructure and database migrations applied through CI.
Vulnerability disclosure
We welcome responsible disclosure. If you believe you have found a security vulnerability, please email security@mnki.com with details and steps to reproduce. Please give us a reasonable opportunity to investigate and remediate before any public disclosure, and do not access or modify data that is not yours.
Incident response
We maintain processes to detect, investigate, and respond to security incidents, and will notify affected users and authorities where required by law and without undue delay.
Contact
Security questions or reports: security@mnki.com.