AI-enabled developer productivity, build systems, and engineering leadership
Posted on in Programming
Retire an API version only after identifying its consumers, verifying replacement behavior, and setting removal criteria that account for clients you cannot deploy yourself.
Posted on in Programming
Give each CI job only the repository, artifact, and deployment authority its work requires, then review that budget when the workflow changes.
Posted on in Programming
Give temporary feature flags an accountable owner, evidence-based removal trigger, and verified retirement sequence before their release purpose ends.
Posted on in Programming
Evaluate dependency-aware test selection with explicit coverage assumptions, full-suite fallback rules, and evidence from shadow runs before trusting it as a CI gate.
Posted on in Programming
Test whether your previous release can read and write changed production state before treating a code rollback as a recovery plan.
Posted on in Programming
Set dependency-update automation boundaries using release risk, meaningful verification, ownership, and a workable recovery path.
Posted on in Programming
Decide when rerunning a failed CI job supplies useful evidence, when it hides an unreliable test, and how to keep both outcomes visible.
Posted on in Programming
Treat slow local tests as a developer workflow investment by measuring iteration cost, choosing the right intervention, and assigning an owner to the result.
Posted on in Programming
Build fast local checks, explicit CI contracts, and clear escalation paths so AI coding agents can make small changes with evidence instead of optimistic guesses.
Posted on in Programming
Give AI-assisted changes a compact handoff contract that preserves evidence, makes the review boundary clear, and leaves consequential decisions with an accountable human.
Posted on in Programming
Investigate a production issue or unfamiliar repository from the terminal without losing the commands, observations, and uncertainty a teammate needs to reproduce your reasoning.
Posted on in Programming
Turn an AI coding agent's reconnaissance into a reviewable engineering plan by separating evidence from inference, naming decisions and risks, and defining validation before implementation begins.
Posted on in Programming
Separate terminal rendering, durable sessions, and remote access so a context switch, disconnect, or laptop change does not turn active engineering work into a scavenger hunt.
Posted on in Programming
An AI coding agent does not need a clever prompt. It needs an assignment that a reviewer could recognize after the fact.
That distinction matters. “Fix the flaky test” may be enough for a human who already knows the service, its constraints, and the team’s habits. To an agent …
Posted on in Programming
Most build tool wrappers start as kindness.
Someone notices that the "real" command is too long, too easy to forget, or too different between local development and CI. They add a Make target, a justfile recipe, a shell script, a package-manager alias, or a small Python helper. Now the team …
