What Mocks Are Actually For
Part 1 of Getting Started with Mocks: the one job a mock does, the five situations where you need one, and what a green mocked test does not prove.
Posts by Matt LeRay
Co-founder and CTO of Speedscale, expert in Agentic AI and cloud data warehousing. • 9 posts published
When Handwritten Mocks Stop Scaling
Part 9 of Getting Started with Mocks: a decision framework for stubs, fakes, fake servers, mock servers, record-and-replay, and real dependencies.
Your First Useful Mock
Part 2 of Getting Started with Mocks: a test you cannot write, one seam, and the smallest mock that makes a delayed package happen. Java, Node, Go, Python.
Your Mock Is Lying
Part 8 of Getting Started with Mocks: the carrier renames a field, production breaks, and every mocked test still passes. Java, Node, Go, Python.
The Compounding Platform Tax
Security controls rarely break a banking platform alone. Their cumulative divergence from Kubernetes standards creates a compounding maintenance tax.
Application Level Dependency Chaos Testing
A scoped chaos rule takes one dependency down and nothing else, so the fallback branch in your code executes on demand instead of during an outage.
Break One Dependency, Not The Whole Cluster
Scoped chaos rules are now in proxymock: pick traffic with a filter query, perturb it, and label every injected failure so it is never mistaken for a real one.
Reliability Engineering in the AI Era
See why reliability engineering must span code, testing, telemetry, and incidents as AI agents erase the boundary between pre-production and production.
From Telemetry to Traffic
Compare eBPF, sidecars, telemetry agents, OpenTelemetry, APM, and proxies through an evolving ladder of production visibility.
Try Speedscale free
Choose the desktop proxymock or the hosted cloud trial to get started.