Tray.ai Launches Agent Gateway for Secure MCP Governance
Gives IT energy to develop accepted MCP Tools on Tray with insurance policies, permissions, versioning and compliance, then publish them by way of MCP for safe agent use throughout the stack
Tray.ai, the platform for constructing good, safe AI brokers at scale, right this moment introduced Agent Gateway, a brand new functionality within the Tray AI Orchestration platform. IT groups use Agent Gateway to construct ruled, maintainable MCP (Model Context Protocol) servers and MCP instruments on Tray and publish them by way of MCP for steady and safe agent use throughout the stack.
As enterprises race to increase brokers with new capabilities, shadow MCP servers and instruments are being developed in JavaScript, Python and with ad-hoc scripts and providers, usually with out IT visibility or required guardrails. As Gartner® famous in its September 2025 report, Innovation Insight: MCP Gateways1, “Enterprises are left attempting to steadiness the keenness for adopting MCP, at the price of governance, towards a extra risk-averse strategy to adoption, on the danger of lacking the innovation boat.”
Tray Agent Gateway supplies a managed surroundings for creating MCP servers and instruments with outlined insurance policies, permissions and versioning, so IT can keep safety and compliance, cut back shadow MCP improvement and put together the agent ecosystem for rising requirements like A2A (Agent-to-Agent).
“Unmanaged MCP use seems quite a bit just like the API sprawl we noticed within the early days of the cloud period,” stated Alistair Russell, Co-founder and CTO of Tray.ai. “API administration centralized and enforced governance throughout APIs together with insurance policies, permissions and versioning. Tray Agent Gateway does the identical for MCP. Now, builders and enterprise groups can rapidly construct MCP instruments utilizing a unified IDE for transparency, collaboration and upkeep. With Agent Gateway and our Enterprise Core working collectively, MCP servers and instruments are all ruled, instrumented and versioned for most management.”
Tray Agent Gateway delivers governance for evolving AI requirements
Agent Gateway supplies the interoperability for MCP and different rising protocols that IT groups use to convey rigor, governance and observability to the enterprise agent stack. Teams can outline, take a look at, model and doc MCP providers, apply scope and guardrails and share them selectively throughout their agent panorama.
Three core methods to make use of MCP are supported by Agent Gateway:
- Build composite MCP instruments: Teams can create refined composite instruments in Merlin Agent Builder and publish them as MCP providers. These instruments can carry out complicated duties or total processes end-to-end and connect with and take motion throughout any system, from apps to different AI providers. Tray Guardian is used to embed guardrails, and customers can mitigate unpredictable agent conduct by encoding enterprise guidelines for constant, auditable outcomes.
- Publish connector-backed MCP instruments: Tray’s library of 700+ managed connectors can now be printed as MCP instruments, immediately giving MCP-enabled brokers safe attain throughout CRM, ERP, HR, analytics and different apps, all with the governance enterprises count on.
- Consume exterior MCP servers: Tray Agents can securely eat exterior MCP servers whereas IT good points centralized visibility, logging and auditability.
“AI brokers will solely scale when enterprises standardize how capabilities are added and ruled,” stated Rich Waldron, Co-founder and CEO of Tray.ai. “Tray Agent Gateway provides IT a single method to set coverage, permissions and versioning so groups can reuse trusted instruments as an alternative of rebuilding them. The result’s decrease danger, decrease price and an working mannequin that may adapt as requirements like MCP and A2A mature.”
Centralized governance and management at enterprise scale
Agent Gateway provides IT centralized command over Tray-built MCP servers and instruments developed throughout the group. By defining MCP servers inside Tray Workspaces and Projects, groups can determine which MCP instruments are uncovered and underneath what situations, so every MCP deployment aligns with enterprise insurance policies and safety scope.
Every MCP software and execution is instrumented in Tray Insights Hub, logged and versioned for traceability. This creates an entire audit path that may be streamed to observability platforms similar to Datadog or Splunk, giving IT clear oversight with out slowing supply. With clearly outlined possession, model management and documentation, IT can keep governance throughout a whole bunch of brokers and instruments, changing fragmented, advert hoc improvement with a single managed surroundings.
Together, Tray Agent Hub and Tray Agent Gateway are a part of Merlin Agent Builder, making a single surroundings for groups to construct, govern and scale enterprise brokers. Agent Hub supplies the composable constructing blocks for agent creation, whereas Agent Gateway delivers governance and management over how these instruments are developed, shared and maintained. With each capabilities in a single expertise, Merlin Agent Builder is now the one-stop surroundings for designing, deploying and managing brokers securely at enterprise scale.
Future-proof structure for multi-agent interoperability
Agent Gateway is constructed as a multi-protocol functionality, supporting MCP right this moment and engineered for Google’s A2A and future agent-to-agent interoperability requirements. With this structure, Tray Agents can talk, delegate and collaborate with third-party brokers whereas sustaining enterprise consistency and management.
By unifying governance throughout protocols, Agent Gateway future-proofs the enterprise tech stack, giving groups confidence that their AI orchestration technique will stay appropriate as new requirements emerge.
“Agentic AI and MCP signify revolutionary know-how, however they’ve additionally launched a brand new wave of technical sprawl,” stated Marcus Dubreuil, Director of Systems Architecture at a world music retailer. “We can now use Tray.ai to spin up new MCP instruments seamlessly and on a managed path. IT approves what’s constructed, enforces insurance policies and permissions, variations it and publishes it for safe agent use. From a single platform, we get centralized visibility and auditability throughout the MCP servers and instruments we deploy—decreasing danger from unvetted code with out slowing supply.”
The put up Tray.ai Launches Agent Gateway for Secure MCP Governance first appeared on AI-Tech Park.
