One of the challenges with using AI and LLMs to generate code today is that they mostly produce generic code. That shouldn't surprise us.
These systems are probabilistic word generators (sorry, no offense to my future AI overlords) trained on a vast ocean of open source projects. What you get is the statistical average of that code — competent, but rarely inspired.
But Model Context Protocol (MCP) changes that game entirely. And if history is any guide, the shift could rival the biggest turning points in how software gets built.
First, What MCP Is (in Plain Terms)
MCP is a way for tools to advertise specialized capabilities to an AI model via APIs.
Instead of treating the LLM as a closed-text generator, MCP lets external systems expose functions ("capabilities") the model can call — securely, with schemas and policies — so generation and validation can be composed.
Think of it like this:
-
Capability registration: Tools (SCA scanners, SAST engines, test runners, refactoring services, linters, build systems, ticketing, etc.) publish what they can do — functions, input/output schemas, and constraints.
-
Discovery and selection: The AI sees an indexed catalog of capabilities relevant to the user's task, along with usage affordances (parameters, costs, scopes).
-
Policy and permissions: Calls are gated by organizational policies (who can call what, with which data), environment scopes, and auditing. Sensitive actions require elevated consent.
-
Invocation and streaming: The AI composes calls (often in parallel), streams intermediate results, and uses outputs to steer the next step (e.g., "SAST flagged X, propose refactor, rerun tests").
-
Observability and feedback: Every call is recorded. Results (pass/fail, severity, timing) feed back into prompts and org analytics for continuous improvement.
-
Decoupled runtime: Tools can run anywhere (local, VPC, SaaS). MCP is the contract between the AI and your stack — vendor-neutral and swappable.
Essentially, MCP turns the LLM from a generalist guesser into a workflow orchestrator that composes specialist tools, your tools, inside the loop.
Echoes of Past Shifts
Software engineering has always advanced through integration moments, times when fractured practices suddenly converged into a new way of working:
-
The IDE Era: Before IDEs, developers juggled editors, compilers, and debuggers. IDEs collapsed the workflow into one place and redefined "writing code."
-
Version Control Systems: Git and GitHub didn't just track changes, they rewired collaboration and made global, distributed development the default.
-
The DevOps Revolution: CI/CD pipelines stitched together testing, builds, and deployments, finally closing the gap between development and operations.
Each of these wasn't just a tooling upgrade. They were discipline enforcers. They didn't just make things easier — they made best practices the default.
The Half-Century of Hard-Earned Best Practices
Today's SDLC is built on the hard lessons of decades past:
-
Software Composition Analysis (SCA): Knowing what's in your code.
-
Static and Dynamic Analysis (SAST/DAST): Catching flaws before attackers do.
-
Unit and Integration Testing: proving correctness at every level.
-
Refactoring and Linters: Codifying lessons about readability, maintainability, and style.
The trouble is, organizations still struggle to get consistency. Some developers embrace every tool, others skip steps, and the result is uneven quality across teams and codebases.
MCP as the Great Normalizer
Here's where MCP's real potential lies. By plugging these capabilities directly into the AI loop, you don't just get smarter code generation. You get consistency.
When AI agents invoke SCA, SAST, DAST, tests, and linters automatically, those practices stop being optional. Every AI-assisted code path follows the same guardrails. And because developers interact with the AI as their primary interface, those guardrails shape their workflows too.
MCP doesn't just upgrade the machine. It levels the playing field across the entire developer base.
Looking Five Years Ahead
If IDEs unified development and DevOps unified delivery, MCP has the potential to unify discipline. Imagine the developer experience in five years:
-
SCA as default: Every new dependency suggested by the AI is automatically scanned for vulnerabilities and license issues before it ever enters your repository.
-
SAST/DAST always on: The AI runs static and dynamic checks in real time, flagging insecure patterns or dangerous flows as the code is generated.
-
Testing as part of creation: Unit and integration tests aren't an afterthought. The AI generates, executes, and validates them automatically. Code without passing tests doesn't even surface.
-
Linters and refactoring by design: Every snippet the AI proposes arrives already styled, linted, and refactored according to your organization's standards.
From the developer's perspective, it feels seamless — just a conversation with the AI assistant. But under the hood, decades of best practices are being enforced uniformly, across every developer, every team, every codebase.
.jpg?width=150&height=150&name=fox-2016-1-sq%20(1).jpg)
Brian Fox, CTO and co-founder of Sonatype, is a Governing Board Member for the Open Source Security Foundation (OpenSSF), a Governing Board Member for the Fintech Open Source Foundation (FINOS), a member of the Monetary Authority of Singapore Cyber and Technology Resilience Experts (CTREX) Panel, a ...
Explore All Posts by Brian FoxTags
Build Smarter with AI and ML.
Take control of your AI/ML usage with visibility, policy enforcement, and regulatory compliance.