Skip to main content

The Future of Agentic Workflows Starts with MCP Intent-Based Server

· 3 min read
Adrian Escutia
Adrian Escutia
A Rebel with a Cause, Innovating the Future

🔥 What if managing AI agents was as simple as deploying Kubernetes?

We're launching the Alpha version and inviting early adopters to shape the next evolution in context-driven AI infrastructure.

MCP Intent-Based Server: Simplifying Context for Complex AI Integrations

MCP Intent Based Server & Tools MCP ServerIntent-Based Managementserver.yamlKubernetes ModelManifest-DrivenReconciliation LoopTemporal WorkflowtoolsPathsModular OrganizationJava-like ReflectionDynamic InstancesDynamicTool CreationDynamicTool LoadingCore ComponentsPrinciplesToolsDesired StateActual StateObserveActAnalyzeFeedbackCorrectionControl Loop

💡 Why This Matters​

Let's face it—enterprise AI integrations are messy. Context gets lost. Tools don't talk. Developers waste hours wiring things together, only for it to break silently in production.

Agentico is building a new foundation, not another abstraction.

We've launched the MCP Intent-Based Server (Alpha):
A dynamic, declarative server for managing the tools, transports, and context dynamically in AI systems—driven by one simple truth:

🧠 The intent should drive the system—not the other way around.

🔍 What's Under the Hood?​

Our alpha release is powered by 3 breakthrough concepts:

1️⃣ Kubernetes-style Manifests
Declare what you want (tools, prompts, connections) in a simple server.yaml. The server reconciles actual vs desired state—automatically.

2️⃣ Temporal-like Paths for Tools
Structure your workspace like Temporal activities and workflows. Just define toolsPaths, and the server loads, manages, and adapts to what you define—dynamically.

3️⃣ Java Reflection-like Tool Injection
Create new tools on the fly from tool names in your manifest. No need to touch core server code. It's plug-and-play for AI capabilities.

❓If Kubernetes abstracted compute through intent, what would it look like to do the same for MCP Servers?

We're betting this is it.

🎯 Who Is This For?​

Whether you're:

  • A platform engineer building internal AI services,
  • A data scientist wanting plug-and-play tooling,
  • Or a startup pushing toward agentic systems in production,

This toolkit is your playground.

We've abstracted the chaos, not the control.

🚀 Alpha Features You Can Use Today​

  • ☑️ Declarative server setup via server.yaml
  • ☑️ Dynamic tool creation & registration
  • ☑️ Transport flexibility (stdio, HTTP, custom)
  • ☑️ Support both MCP and standard server types. (The McpServer provides access to the server, but read-only access)
  • ☑️ Tool lifecycle management: register/unregister on manifest change
  • ☑️ Built-in logging for debugging

📬 Want Early Access?​

Join the alpha program, contribute ideas, and shape the future of Model Context Protocol in real enterprise scenarios.

🛠 GitHub Repo with example
📩 Subscribe for updates
🎥 Follow us on LinkedIn and Adrian's

🧢 La Rebelion Merch.

📢 Our different blogs:​

  • La Rebelion - Topics around Cloud Native, Kubernetes, and DevOps.
  • Agentico - Topics around Agentic workflows, MCP, and AI.
  • Intent Based AI - The specs and the tools to help you with Intent Based AI.
  • MCP - Our hub for the Model Context Protocol (MCP) and the tools we are building around it.

🧭 Looking Ahead​

We're already building:

  • Multi-server support for scale-out deployments
  • Telemetry, observability & versioning
  • Native gRPC + custom transports
  • Integrated rollback strategies

This is just Day 1.

We want to hear from you. What do you need? What are your pain points? How can we make this even better?

Go Rebels! ✊🏽