I'm Bruno Souza, an integration architect. I work with MuleSoft, APIs and CI/CD. Outside work, I study how to use AI agents in software development.
Scatter-Gather, Choice, For Each and the Anypoint MQ dead letter queue come from the book Enterprise Integration Patterns. Knowing the name of the pattern shows the problem the component solves, what it costs and when you can do without it.
A folder in your vault becomes the blog. The note's name becomes the address, the top of the note becomes the page, and the site refuses anything that is wrong. Free, with a ready-made Astro template.
A short request, a Mule 4 API and OpenSpec show how a spec reviewed before any code reveals what the request forgot and catches a breaking change.
What the research says about AI and software development, and why I think fundamentals and review are still the job.
A plausible 200 response, an inactive customer and the rule missing from the requirement.
Retry alone turns a degraded backend into an integration outage. A public, measured, tested plugin shows the difference — and what changes once you add replicas on CloudHub 2.0.
Path: SOLID in MuleSoft · Part 6 of 6
DIP is not externalised configuration. It is the business flow defining the model it needs and adapters conforming to it — shown in Mule 4 with swappable ERP and in-memory implementations.
Path: SOLID in MuleSoft · Part 5 of 6
One order endpoint that returns everything to everyone, split into consumer-shaped interfaces — and why the strongest argument for ISP in integration is data exposure.
Path: SOLID in MuleSoft · Part 4 of 6
Mule has no inheritance, but LSP still bites: two adapters behind one contract, where the new one keeps the field names and quietly changes their meaning.
Path: Delivering MuleSoft applications with Azure DevOps · Part 4 of 4
Promote the same Exchange artifact through test, uat and prod on CloudHub 2.0, with approvals on Azure DevOps environments and a smoke test after every deployment.
Path: Delivering MuleSoft applications with Azure DevOps · Part 1 of 4
Why GitFlow still fits scheduled MuleSoft releases, where it doesn't, and how to turn branch names into pipeline behaviour — with a working reference project.
Path: Delivering MuleSoft applications with Azure DevOps · Part 2 of 4
release:prepare and release:perform explained step by step, running inside Azure Pipelines — detached HEAD, pushing with the pipeline token, [skip ci] and branch protection.
Path: Delivering MuleSoft applications with Azure DevOps · Part 3 of 4
How to structure versioned Azure Pipelines templates for Mule applications — stages, jobs and steps, secrets without secure files, and why MUnit is left out of a public pipeline.
Path: SOLID in MuleSoft · Part 3 of 6
Adding a payment method to a Mule 4 app without touching the router — with config-driven dynamic routing, MUnit evidence, and an honest look at the trade-offs.
Path: SOLID in MuleSoft · Part 1 of 6
A code-backed look at the five SOLID principles in Mule 4 — where each one fits integration work, where it needs translating, and where it stops being useful.
Path: SOLID in MuleSoft · Part 2 of 6
SRP is not about small flows. It is about grouping logic by the reason it changes — shown with a Mule 4 order flow before and after, with MUnit tests.
