Skip to content
OverviewCapabilitiesHow it worksMCP CatalogSecurityPricing
Login ↗Register ↗
Login ↗Register ↗

LEGAL & TRUST

OverviewPrivacy PolicyTerms of UseCookie PolicyAcceptable Use PolicySecurity & DisclosureAccessibility Statement
contact@aphura.ai ↗

APHURA / LEGAL & TRUST

Security & Disclosure.

Security boundaries, reporting guidance, and the current qualification stage.

Last updated October 19, 2026

ON THIS PAGE

1. Security approach2. Important boundaries3. Reporting a suspected vulnerability4. Research boundaries5. Coordination and incidents

1. Security approach

Aphura combines verified identities, tenant-scoped access, signed policy, exact-operation human approvals, action budgets, and tamper-evident audit evidence. Its customer-local MCP gateway inspects traffic and reviewed tool definitions before authorizing calls.

These are descriptions of implemented controls, not a guarantee that every threat is prevented or that every deployment is secure.

2. Important boundaries

Controls apply to actions routed through Aphura. Deployment must prevent agents from bypassing the gateway or obtaining its upstream credentials. A gateway outcome records what the gateway attests it observed. Platform suspension does not itself terminate every workload, revoke external identity-provider or cloud credentials, or reverse an action already performed.

Aphura is in development. Local integration and browser checks are not substitutes for target-environment validation, independent security assessment, or customer acceptance. No compliance certification, production SLA, or guaranteed recovery target is asserted here.

3. Reporting a suspected vulnerability

Email contact@aphura.ai with the subject Security report. First provide a non-sensitive summary and ask for an appropriate secure transfer method if the issue requires confidential evidence. Do not include production credentials or customer data in ordinary email.

Include a concise description, the affected component or version, the expected versus observed behavior, and minimal reproduction steps using synthetic data. Remove credentials, customer information, and other unnecessary sensitive material.

4. Research boundaries

Do not access another tenant's data, use stolen credentials, persist in an environment, attempt social engineering, or perform availability-impacting tests. Stop testing if you encounter personal information or unexpected access and preserve only the minimum non-sensitive evidence needed to explain the issue.

Testing requires permission from the owner of the specific environment. This page does not grant general authorization, promise a bounty, set a response deadline, or establish a legal safe harbor.

5. Coordination and incidents

Use a verified private channel for a potential vulnerability rather than posting exploit details publicly. Coordinate any further access, evidence handling, and disclosure with the affected operator. A confirmed customer incident follows the applicable customer agreement and incident-response process, rather than relying solely on a general website enquiry.

All legal documents ↗Contact Aphura ↗

The control layer for AI agents.

Back to top ↑
© 2026 Aphura, Inc.
PlatformHow it worksSecurityPricingArchitectureFAQContact
In development.
Legal & trust
Privacy PolicyTerms of UseCookie PolicyAcceptable Use PolicySecurity & DisclosureAccessibility Statement