Clear thinking about the Model Context Protocol: limitations, misconceptions, and practical patterns for building reliable agent systems. Learn from deep dives and hands-on guidance.
We see HAPI MCP as the convergence layer between OpenAPI (OAS), MCP, Arazzo, and LLMs — “OAS v4” for AI and agentic workflows. It separates design from runtime, bringing modularity, reliability, and observability to AI systems.
Separate design artifacts (prompts, tools, policies) from runtime execution for faster iteration.
Standardized interfaces enable interoperable agents and services across teams and infra.
Deterministic contracts with validation make agent behavior predictable and testable.
Inspect flows, events, and metrics to continuously improve quality and safety.
Principle-of-least-privilege for tools and data access with clear boundaries.
OAS + MCP + Arazzo + LLMs provide a portable, vendor-neutral foundation.
Assist agents, triage requests, and automate resolutions with grounded context and safe tool use.
Unify knowledge graphs and services for accurate, explainable retrieval-augmented workflows.
Governed analysis pipelines with human-in-the-loop approvals and reproducible outputs.
“MCP gave us predictable agent behavior and faster releases.”
“The design/runtime split made compliance sign‑off straightforward.”
“We moved from prototypes to production with clear contracts.”
No. We see MCP as complementary — it brings agent context and tool protocols. Together with OAS and Arazzo, you get a cohesive contract for agentic workflows.
HAPI MCP implements the protocol and adds production‑ready features: auth, routing, orchestration, and observability across services and agents.
Yes. The interfaces are model‑agnostic. You can bring your preferred providers and switch without changing integration contracts.
Absolutely. MCP builds on service boundaries. Wrap capabilities as tools, describe contracts, and compose workflows safely.
Use scoped credentials, audited tool calls, and least‑privilege access. HAPI MCP adds guardrails and observability for regulated environments.
Insights on MCP, patterns, pitfalls, and hands-on examples delivered monthly.
See HAPI MCP in action — the convergence of OAS, MCP, Arazzo, and LLMs.
An evolution of tool calling with MCP, thanks to OpenAI's latest SDK updates.
Step-by-step, I'll guide you through setting up an MCP server, integrating it with the OpenAI SDK, and running a complete example that showcases dynamic tool calling. By the end of this post, you'll be equipped to leverage MCP in your own OpenAI-powered applications.
End-to-End Example, Setting Up an MCP Server, Integrating with OpenAI LLM, and Running some tests to see it in action.
Prompt injection is a critical security risk for any system using large language models (LLMs), including those built with Model Context Protocol (MCP). You must understand how prompt injection works, why MCP cannot prevent it, and what steps you should take to protect your users and applications (MCP Clients).
🚨 "OpenAPI Specification (OAS) v4 is out" - That I wish, this is the kind of headline I would expect to see soon, because OAS can easily be extended to enable RESTful APIs work seamlessly with AI.
By the end of this article, you'll know how to let any LLM call your REST tools automatically using OAS.
