GitKraken

Every time your AI keeps missing the mark on a feature, there's a good chance you needed a spec, not a better prompt. Erik Hanchett, Senior Developer Advocate at Amazon Web Services (AWS), introduced a framework we've been thinking about ever since: spec or vibe? Vibe coding is great for small, fast changes. But the moment a feature gets complex, the moment your PM hands you a list, the moment you find yourself going back and forth with your AI and getting nowhere... that's the signal. That feature needed a plan before it needed a prompt. Erik walked through a full spec-driven development workflow live, showing how a structured requirements doc, design doc, and task list gives AI coding agents the context they actually need to build what you asked for. The full talk is on our YouTube channel. Worth 16 minutes of your time if you're using any AI coding assistant seriously. https://lnkd.in/gJwDkrxX #AIcoding #DeveloperTools

We've all been there. Forty commits deep. Half of them say "fix." Two say "fix 2." One just says "ugh." Your Git history is a crime scene, and code review is the detective. Commit Composer is our answer to that. It uses AI to take whatever mess is sitting in your working directory, or across a full commit range, and turn it into clean, logical commits with messages that actually mean something. No rebase gymnastics. No cherry-pick limbo. Your original branch stays untouched until you're ready to commit. When it's done, your history looks like you had a plan the whole time. Because the code was always good. The commit messages just didn't show it. Try Commit Composer in GitKraken Desktop & GitLens.

Merge. Rebase. Squash. Three ways to integrate changes, three very different commit histories. Knowing when to use each one separates a clean repo from a chaotic one. Here's the quick breakdown. There's no universally "right" answer. The right choice depends on your team's workflow, your PR process, and how much history matters to you. What's your team's go-to? Drop it in the comments. #Git #GitKraken

Leadership isn't asking "are we using AI?" They're asking "what did it actually cost us, and what did we get?" Those are two completely different questions. And most engineering teams are still answering the first one when the second is the only one that matters. Our VP of Engineering, Anastasia Zamyshlyaeva, is speaking at LeadDev on how to build a metrics framework that connects AI adoption to the outcomes boards and CFOs actually care about: code quality, developer satisfaction, and business performance. Panelists include senior engineering leaders from Phrase and Quotient, with Spotify's Engineering Manager moderating the conversation. Register: https://lnkd.in/g2XaM_DV #EngineeringLeadership #AIROi #LeadDev

Something I really like about GitKraken's AI generated commits is the ability to tune the prompt that is used. let you tweak the generated output to suit your liking. It saves me tons of time each week. https://lnkd.in/gBka3GhQ

This is absolutely true, especially when summarising multi-file changes! Establishing guidelines for different types of commits makes the process even easier. GitKraken has significantly sped up these moments that I used to dread.