When we launched MCP Bridge, it gave teams a governed path to turn existing APIs into agent-ready tools without writing MCP server code from scratch. That solved the build problem. Two problems remained. First, the long line of SaaS systems agents needing to act on CRM records, project tasks, support tickets, etc. still required separate integration tracks. Second, once an MCP server was deployed, any meaningful change meant rebuilding it from scratch and migrating every agent that depended on it. This release addresses both.
We heard your problems and took action
These friction points come up in our conversations with customers:
- “We built an MCP server in-house to expose a few internal APIs. It took three engineers six weeks. Now we need to update it and we’re afraid to touch it.” – Enterprise Integration Lead
- “We want to give our agents access to core enterprise and other SaaS tools alongside our own APIs. Right now those are completely separate integration tracks.” – AI Platform Architect
- “Every time we update the underlying API, we have to rebuild the MCP server. It’s becoming a bottleneck for the whole agent program.” – Platform Engineering Manager
Separate tracks for SaaS and custom APIs. Rebuilding MCP servers on API update. Teams avoiding changes rather than dealing with the disruption. These are not the edge cases, these are the norms for teams moving agent programs from experimentation into production.
For teams building towards a headless architecture where backend capabilities are exposed as APIs and MCP tools and consumed by whatever surface needs them, human or agent, this brittleness is a direct blocker.
3 steps, more capabilities with MCP Bridge
MCP Bridge keeps the same three step creation flow from the original launch. What’s changed is what you can include in each step and what you can do once a server is deployed.
- Select your sources: Choose from your registered or discovered APIs, or pick from 100+ prebuilt SaaS actions covering Asana, Jira, Mailchimp and more. SaaS actions and custom APIs can be combined into a single server. What this means is an agent can query a Zendesk ticket, check the related Asana project and call an internal order management API from one governed endpoint.
- Configure your tools: Select which API methods or SaaS actions to expose then set tool names, descriptions, and schemas, the metadata agents use to understand and invoke capabilities correctly. Inline schema editing lets you make adjustments without leaving the flow
- Deploy: MCP Bridge generates a governed MCP server ready for production. Auto policy configuration detects common policies – OAuth, rate limiting, client credentials – and pre-configures them inline as recommendations, so governance is baked in without manual setup. No infrastructure to manage. Every server is automatically published to Anypoint Exchange as a versioned, discoverable asset alongside your existing APIs
SaaS actions ready to activate
The hardest part for enterprise agent programs isn’t internal APIs – it’s the SaaS systems that agents need to act on. Before this release connecting an agent to Jira, or Zendesk meant a separate integration track with its own build and maintenance overhead.
MCP Bridge now ships with 100+ natively supported SaaS actions across systems including Asana, Mailchimp, HubSpot, Jira, and more. Select the SaaS systems and actions they want to expose directly in the MCP creation flow, combine them with your own APIs and manage everything from one governed endpoint.
| “We spent weeks wiring up existing APIs and a few other tools for our agents. If we could have just selected them from a list and mixed in our own APIs, we’d have shipped months earlier.” – AI Product Lead |
Update and version without breaking your agents
Previously, any meaningful change to a deployed MCP server updating a tool description, adding a new API method, swapping to a newer API version, required creating a new server and migrating all agent integrations. Now, changes to a deployed MCP server generate a new underlying MCP spec – a versioned record of what changed and when – without changing the MCP server’s URL or disrupting the agents connected to it.
What you can do now after deployment:
- Edit tool names, descriptions, and input schemas without redeployment
- Add new API methods as additional tools and publish as a versioned release
- Swap the underlying API to a newer version without changing the agent-facing contract
- Add new APIs or SaaS systems to an existing MCP server
- Start with a few tools and expand incrementally as agent use cases mature
The approach is the same one teams already use for managing API lifecycle. publish a new version, deprecate the old one on your timeline, let consumers migrate on a schedule you control. Several improvements reduce the operational overhead of getting an MCP server from prototype to production.
One catalog for humans and agents
Every MCP server created in MCP Bridge is automatically published to Anypoint Exchange as a cataloged, documented asset – versioned, discoverable, and available for reuse across teams.
API Experience Hub extends this to the developer-facing portal. Internal and external developers now see a single portal that surfaces REST APIs, MCP servers, AI agents, GraphQL, and gRPC side by side – alongside APIs federated from other gateways.
The significance here goes beyond developer convenience. Enterprises are increasingly architecting around the headless principle, where the same API layer powering human-facing digital experiences should also power agent-facing automation – a single composable foundation that serves both. Customer experience is what agents do on their behalf. But that only works if human developers building frontends and autonomous agents executing workflows are working from the same catalog, with the same governance, and the same visibility. MCP Bridge makes that concrete.
The APIs your teams have already built and governed don’t need to be re-exposed, re-documented, or re-managed for the agent layer. They extend naturally into it. The portal your teams already use for API discovery becomes the front door for your entire agent capability catalog – no separate experience for AI tools versus integration assets, no diverging governance models.
Getting started
Ready to simplify your API-to-agent journey? Here’s what you can do now:
- Explore the documentation: Dive deeper into the features, policy configurations, and full lifecycle management of MCP Bridge by reading the full documentation
- Create your first MCP Server Log in to MuleSoft and start creating your first production-ready MCP server in three simple steps
- Watch the demo video to learn even more