TRAPDEFENSE / CLOUD & SUPPORT DIRECTION

Open source first.
Operations when
you need them.

TrapDefense 0.47 publishes the Runtime Gateway and Agent Access Broker together under MIT. There is no separate Enterprise code edition. You can self-host the complete current product; future commercial value can come from running, integrating, and supporting it.

THE SAME PRODUCT BOUNDARY
RUNTIME GATEWAY + CONTENT POLICY + AGENT ACCESS BROKER + AUDIT

The enforcement path stays inspectable and self-hostable. A future service would operate the same product contract rather than unlock hidden authorization features.

POSSIBLE COMMERCIAL SERVICES

Sell the hard operational work,
not a code gate.

These are direction statements. They are not launched offerings, published SLAs, or promises of current availability.

01 / MANAGED CLOUD

Operate the boundary.

Managed deployment, upgrades, health, backups, tenant isolation, and supported private connectivity for compatible traffic paths.

02 / FLEET & EVIDENCE

Scale operations.

Central policy rollout, multi-node HA, capacity management, immutable external audit, and evidence retention integrations.

03 / INTEGRATION & SUPPORT

Fit the customer stack.

Identity mapping, MCP/API routing, destination credentials, policy design, acceptance testing, rollout support, and incident response.

WHAT IS AVAILABLE NOW

The open-source 0.47 baseline.

Access Broker is Experimental. Customer IdP policy, production routing, target authentication, bypass prevention, HA, capacity, and operational ownership still require deployment acceptance.

No hosted service is available today.

There is currently no managed signup, subscription, production SLA, or multi-tenant control plane. Use the open-source self-hosted package and contact us only if the explicit integration contract fits your environment.

Review the deployment contract ↗
START WITH DEPLOYMENT FIT

Bring one
real action path.

A useful first engagement defines one client, one destination, a small tool/action set, identity claims, failure policy, and measurable acceptance tests.

hellocosmos@gmail.com ↗
GOOD FIRST SCOPE

One protected workflow

Configurable client endpoint · bounded HTTP/MCP JSON · local agent credential, external issuer or connection key · separate target credential · explicit allow/deny/approval cases · bypass control · sanitized evidence.

Discuss deployment fit ↗