GitKraken

If attackers can automate vulnerability hunting with AI, so can defenders. Great take from Chris:

We didn’t have a tool to measure whether our own devs were actually using AI. So we built one. Then we built another to act on it. The result: 3.5x engineering productivity and ~97 hours unlocked per dev. No bigger budget. Our own case study: gitkraken.com/customers/gi...

IDE: you write the code, agents help. CLI agent: one agent per terminal. ADE: you assign the work, agents run in parallel, you review. What an agentic development environment actually is: gitkraken.com/blog/underst...

AI is getting really good at the things you are bad at. First video in a long time just dropped.

5 agents running. Which one needs you? Kepler’s Inbox view sorts tasks by what they’re asking of you: unread, blocked, running, done. Your attention goes where it’s needed 👇 help.gitkraken.com/kepler/arran...

@gitlens.bsky.social 19.4: ✅ Commit Graph, no account required ✅ Branch names generated from your issues ✅ Publish repos straight from the Graph ✅ Auto-fetch that stays out of your pulls gitkraken.com/blog/gitlens...

Unpopular opinion: Being a senior engineer has almost nothing to do with writing code. It’s about communication, unblocking your team, and making decisions when there is no "right" answer.

If you've typed the same prompt to an agent twice, it should be a button. Kepler Actions are named prompts you fire at a task, issue, or PR. You can edit the built-ins, write your own, and pin each one to a different model. help.gitkraken.com/kepler/actio...

84% of devs say AI makes them more productive. 39% of their orgs have no way to measure it. Another 33% just ask the devs. New data from 554 devs and engineering leaders: gitkraken.com/reports/stat...

Yesterday, I discovered @gitkraken.com MCP App while asking for something in the chat of VS Code. Really nice

Kepler 0.11 is out, and hitting update no longer kills your agents. Agents and terminal sessions survive restarts and updates, then reconnect when Kepler comes back. Claude artifacts now stay attached to the task. Plus up to 16 Dashboard previews side by side. gitkraken.com/kepler

Only 24% of devs we surveyed can prove their AI agents are paying off. 45% are going on gut feel. 21% can't tell. 8% say they're spending more without moving faster. Here's what else they told us about agent cost and ROI 👇

Before you hand an AI agent your merge conflicts, check how it scores on Git. @cmgriffing.bsky.social's 18-minute tour of GitBench: model comparisons, reasoning levels, token usage. The repo is public, so you can run it against your own models. youtu.be/y7Oug2rNB-s

Running more agents doesn't help if every task still starts with a blank prompt. Tomorrow we're showing how to turn agent workflows that work into reusable patterns with Kepler and Insights for Developers. Register to win a portable monitor: gitkraken.com/webinar/scal...

Running five agents across twelve branches and trying to figure out what's ready to merge? Kepler runs Claude Code, Codex, Copilot, and Cursor in parallel, each in its own worktree, and shows you every agent's work on one graph. gitkraken.com/kepler

@codermatt asked which software you'll keep using no matter how good AI gets, and picked us. Then @0xCaponte said he uses us even more now, to see what his agents change after each step. We'll be framing this one.

Four lines in a CLAUDE.md file can change how an AI assistant works in your codebase. Tell it your review standards, naming conventions, architectural boundaries, and definition of done, and it stops generating code that needs heavy rework before it's review-ready.

I really wasn't expecting to have to create a GitBench v2, yet. But the models are getting really good at git.

The models got too good at our Git benchmark. Time to make it harder. Latest GitBench results: gitkraken.com/blog/models-...

@cmgriffing.bsky.social goes whole days without opening an IDE now, mostly thanks to Kepler. He still reaches for one to review and trim what the agent wrote, because go-to-definition beats prompting an agent into minimal code. His take on whether the IDE is dying: gitkraken.com/blog/challen...

Kepler doesn't make you go find your work. It opens on it: every issue and PR assigned to you, plus the tasks already running, in one list. One click hands an item to an agent with full context attached. help.gitkraken.com/kepler/keple...

Most teams use AI for one thing: writing code. That's maybe 20% of what it can do. Code Flow isn't about writing faster. It's about the whole pipeline moving faster, from generation to review to ship. The next competitive edge isn't AI-assisted writing. It's AI-assisted reviewing. #CodeFlow

Developers will debate for an hour whether AI should write their code. They won't debate whether it should write their commit message. 369,565 commit messages generated in GitKraken in the last 30 days, our most-used AI feature by a wide margin. The tax gets automated. The judgment stays yours.

Still running four terminal tabs just to babysit your agents? GitKraken Desktop 12.5 closes that loop. & when an AI commit blends work from multiple people, it credits all of them instead of pinning it on whoever hit compose. Try 12.5 → help.gitkraken.com/gitkraken-de...

GitLens 19.2 is here! GitLens now speaks your language, and the Commit Graph gets even better for worktree workflows. Here’s what’s new 👇 

Huge thank you to all who came by our booth during LeadDev this week. Our team loved being able to meet so many of you in person and chat GitKraken Insights, Kepler & the future of agentic development. See you at the next one!

I predict ADEs like Kepler will start adopting this pattern themselves. It's way better than trying to support every harness under the sun and people can vibe code their own harnesses and still use them in an ADE like this.

Pi is one of the leanest agent harnesses around. No approval prompts, no hand-holding. It also doesn't support ACP out of the box, so it's not officially in Kepler yet. @cmgriffing.bsky.social found a workaround, and the 2 bugs still standing in the way: gitkraken.com/blog/you-can...

A developer just called us out for launching Kepler too quietly. Fair. We're done being silent. Kepler runs any agent, any model, in parallel, with Agent Graph showing exactly what's happening as it happens. From idea to merged PR, one surface. gitkraken.com/kepler #AgenticDevelopment

GitKraken was a hackathon side project before it was GitKraken. The team was building a tool. Git kept getting in the way. So they fixed it during an internal hackathon, and that fix became the product. @cmgriffing.bsky.social shares why more teams should run these: gitkraken.com/blog/run-mor...

Counterintuitive fix of the week: disable your GPU to make GitKraken run better on WSL. Straight from our support team: 4 things generating tickets right now: gitkraken.com/blog/turn-off-your-gpu-to-fix-gitkraken-on-wsl-and-three-other-things-support-is-fielding-this-week

Someone asked what VS Code extension they couldn't live without. This is exactly why we build GitLens.

Running 3+ AI coding agents at once? The upside grows, but so does the chaos. Join us Sept 17 @ 1p EST to see how GitKraken Insights for Developers + Kepler give you visibility and structure for multi-agent workflows, without you becoming the bottleneck. Register: gitkraken.com/webinar/orch...

We asked devs if they could prove their AI agents are paying off. Only 24% said yes. The rest are just running on gut feelings, guessing, or already suspecting they’re spending more than they’re saving. Full poll data: gitkraken.com/blog/can-you... #CodeFlow #GitKraken

Your Git workflow shouldn’t be creating more work than the code itself. Get GitKraken Desktop, GitLens, and Kepler access together for up to 60% off for just a limited amount of time. gitkraken.com/stop-the-slo...

Not sure what Kepler is or how to use it? We built a full tutorial series covering tasks, actions, the Agent Graph, and running agents in parallel without losing the thread. Watch: www.youtube.com/playlist?lis... Or jump in now: gitkraken.com/kepler #CodeFlow #GitKraken #Kepler

"It's very difficult, at least for me, to parse and focus on what's going on there." Our Director of Engineering on why watching an agent's raw text log doesn't work, and why Kepler now shows you a graph instead. gitkraken.com/blog/why-we-...

We went live to launch GitKraken Insights for Developers. Cost and usage data down to the PR level, plus parallel agent workflows in one surface. Full recap, plus the live Q&A on dev containers, worktree cleanup, and what's next: gitkraken.com/blog/live-la...

GitLens 19.1 is here 🦑 More coding agents in the Graph. Better worktree workflows. Smarter Git maintenance. Fixup commits + autosquash. And a new way to see the health of your repo. Here’s what’s new 👇

Congrats to Oleg Cohen for winning our Insights & Kepler webinar giveaway! Want a shot at the next one? Register for our upcoming webinar and you're automatically entered: gitkraken.com/webinar/orch...

We know you’re tired of the AI hype. It feels like more code, more agents and more noise to sift through. Kepler is our attempt at less noise with every agent session on one board with nothing hidden. Meet Kepler: gitkraken.com/kepler

Companies aren't just shipping AI models anymore, they're shipping them undercover. No name, no press release, just a codename and a community trying to crack who's behind it. @cmgriffing.bsky.social breaks down the Ox Alpha mystery: gitkraken.com/blog/what-ar...

You can run five different AI agents on five different models across your repos. The hard part isn't running them. It's staying in control of what they're doing. Kepler ADE tutorial #1 covers the interface and the core ideas the rest of the series builds on. Start here. youtu.be/U0w18ZXrZ5M

More agents. More models. More guesswork? Join us Monday, Aug 31 @ 1PM EDT for a live look at GitKraken Insights for Developers and Kepler, built to show what's actually working in your agentic workflow. Register for a shot at a PS5. gitkraken.com/webinar/more...

Everyone's calling Perplexity dead. The OpenRouter benchmarks disagree, it won 11 of 12 search matchups using the same models everyone else has. The edge is the search layer, not the model. gitkraken.com/blog/dont-sl...

Ask a developer if AI made them faster this year. 84% say yes. Ask their VP to put a number on it for the board. 39% have nothing to show. That gap, not adoption, is the real story in engineering right now. We surveyed 554 devs and leaders to map it: gitkraken.com/reports/stat...

AI was supposed to mean less work. For a lot of developers, it's meaning more. Fast feedback + visible results + more power = a loop that's genuinely hard to stop. Engineering leaders on a recent panel break down why, and what to watch for: gitkraken.com/blog/ai-was-...

Starting AI agents is easy. Keeping track of them is the hard part. Kepler runs every agent in one place, from idea to merged, every session visible live. Bring your own agents and models. Free in public preview. gitkraken.com/blog/big-kep... #AIagents #DevTools #Kepler