Security Model Overview

Security concerns are one of the first questions teams ask when evaluating a new integration technology.

Graftcode is designed with a simple principle in mind:

Note

Execution always happens inside your environment.

Graftcode never becomes part of your production data path.

This section explains how security and trust boundaries are defined, enforced, and preserved.


Graftcode is not a cloud proxy

The most important clarification is what Graftcode is not.

Graftcode:

  • does not proxy production traffic
  • does not process business data in the cloud
  • does not participate in runtime execution
  • does not observe payloads or method arguments

All execution happens:

  • inside your processes
  • inside your runtimes
  • inside your network boundaries

The Graftcode cloud is used only for interface metadata and tooling, not execution.


Clear trust boundaries

Graftcode defines clear and narrow trust boundaries.

Trusted components (your environment)

  • Application code
  • Business logic
  • Hypertube runtime
  • Graftcode Gateway process
  • Runtimes loaded by the Gateway
  • Transport plugins and security plugins you choose

These components run under your control and follow your security policies.


Untrusted / non-executing components

  • Graftcode cloud services
  • Registry and metadata services
  • Tooling and dashboards

These components:

  • never execute your code
  • never see your runtime data
  • never participate in method calls

They operate strictly at the metadata and orchestration level.


What data is shared with Graftcode cloud

When a Graftcode Gateway starts, it may share interface metadata.

This includes:

  • names of public classes
  • method signatures
  • type information
  • version identifiers

It explicitly does not include:

  • method implementations
  • runtime values
  • business data
  • credentials or secrets

The shared data is comparable to a public header or interface definition.


Execution isolation and safety

Execution isolation is provided by:

  • runtime boundaries
  • process boundaries
  • network boundaries

Hypertube executes code:

  • inside the native runtime
  • using native execution rules
  • without injecting or modifying logic

There is no interpreted execution layer and no sandbox escaping risk introduced by Graftcode itself.


Authentication and authorization are explicit

Graftcode does not implement a built-in authentication system.

Instead:

  • authentication is handled through plugins
  • authorization is enforced explicitly
  • identity flows are transparent

This ensures:

  • no hidden credentials
  • no implicit trust
  • no security logic embedded in transport

Security decisions remain visible and auditable.


Network security is unchanged

From a network perspective, Graftcode introduces no exotic requirements.

Communication uses:

  • TCP/IP
  • or WebSocket

This allows teams to:

  • apply existing firewall rules
  • use standard TLS / WSS
  • rely on existing network monitoring and inspection

Graftcode fits into existing network security models.


Observability without data leakage

Observability data:

  • traces
  • spans
  • logs
  • metrics

remains under your control.

Graftcode:

  • propagates context
  • does not centralize logs
  • does not collect payloads

Teams can integrate with their existing observability stack without exposing sensitive data.


Source availability and auditability

Graftcode components that run in customer environments are designed to be auditable.

Where applicable:

  • components may be source-available
  • behavior can be inspected
  • security reviews and penetration testing are possible

This supports enterprise security review processes.


Failure modes and blast radius

Because execution does not leave your environment:

  • failures remain local
  • blast radius is limited
  • outages in Graftcode cloud do not affect production execution

At worst:

  • new Grafts cannot be generated temporarily
  • existing deployments continue to function normally

This is a deliberate design choice.


Trust through minimalism

The security model of Graftcode is intentionally minimal.

It avoids:

  • deep cloud involvement
  • opaque runtime behavior
  • implicit trust relationships

Instead, it relies on:

  • explicit boundaries
  • standard protocols
  • native runtime execution

This keeps the security surface area small and understandable.


Closing perspective

Graftcode does not ask you to trust a new execution platform.

It asks you to trust:

  • your own runtimes
  • your own network
  • your own security practices

Graftcode stays out of the way—and makes those boundaries clearer, not blurrier.