Kiro
@kirodotdev

Agentic AI development from prototype to production. Kiro helps you do your best work by bringing structure to AI coding with spec-driven development.

kiro.dev
Joined May 2025
252 Photos and videos
Class is dismissed. 📖 This week, we opened the doors to the Kiro University Challenge: a crash course built for builders. What are you building?
11
3
28
2,832
Kiro University Challenge, Day 4. Lesson 7: Powers Install and use a Power so Kiro loads its tools and skills on demand. Share what you're building with #KiroUniversity. spr.ly/6017BG5pTX
12
2
16
3,401
Kiro University Challenge, Day 3. 📚 Lesson 5 & 6: MCP + Custom agents Share what you're building with Kiro University. Syllabus 📖 spr.ly/6012BG5FQn #KiroUniversity
10
3
28
5,155
📚Halfway through the Kiro University Challenge, our one week Kiro crash course. Time for a check in on your build. How's your project coming along? Tell us in the comments 👇 Syllabus:spr.ly/6018BG5pHa #KiroUniversity
11
1
31
4,084
"It's never been easier to build something than it is today." A UT Austin sophomore built a 3D explorer of Austin. Runs right in your browser. Started as an idea in his notes app. He typed it into Kiro, reviewed each step as it built, and kept going. Apply today: spr.ly/6011BGe8QF
14
3
27
4,336
Class is back in session 🎓 The Kiro University Challenge starts today. A one week crash course for Kiro. 📚 7 lessons: steering, hooks, MCP, Powers + more 🛠️ Build a project as you go 🏆 Up to 5,250 credits Syllabus here: spr.ly/6011BGbvT5 #KiroUniversity
4
37
188
14,001
Claude Fable 5.1 is now available to Kiro enterprise customers. It holds context across long-running, context-heavy agentic sessions, and traces problems to the root causes in the code. Learn more: spr.ly/6019BGboP1 Stay up to date with the latest models: spr.ly/6010BGboPG
21
8
98
9,826
Back to school. 📚 New semester, new lessons. Something exciting is coming to Kiro next week. Build as you learn. Stay tuned. Open to everyone. Not just students. #KiroUniversity
7
9
82
8,180
Kiro students just went global 🌎 Expanding to 121 new universities across 16 countries. Same deal for everyone: a full year free, 1,000 credits/month, premium models + Kiro Web included. Sign up and learn more here 👉 spr.ly/6018BGHxls
56
49
647
158,180
Kiro Web + cloud sessions are generally available, with cloud configuration. Build from the browser, or start a new session from the CLI or IDE. Pick up the same session later from your browser, CLI, IDE, mobile, or another machine. What changed 👉 spr.ly/6010BG6kYE
10
16
50
8,354
more replies
With cloud configuration, your steering, custom agents, skills, powers, hooks, and MCP settings travel into cloud sessions alongside project configuration from the repo. Your configuration sync also loads into new local IDE and CLI sessions.
1
2
903
Available now on Pro, Pro+, Pro Max, and Power Uses your existing Kiro credits, with no separate charge for cloud compute. Give Kiro the next thing you're working on → app.kiro.dev
1
817
Have you tried KiroCrew 0.4.0? The dashboard does more than watch. You can work inside it now.
6
5
42
5,522
more replies
Windows and Linux are first-class now: signed installers, desktop automation on Windows, and matching resource ceilings on both platforms.
2
3
922
Every messaging channel catches up to Slack, and WhatsApp, iMessage, and Feishu join the roster too. Full breakdown 👉 spr.ly/6015B1hHaz

Release v0.4.0 · kirodotdev/KiroCrew

Windows and Linux stop being second-class: both get signed, self-updating native installers, and Windows gains desktop automation and a real resource ceiling. Every messaging channel catches up to ...

github.com
625
How do you screen 27 prompt changes in a week without mixing their effects? @kirodotdev assigns developers to stable cohorts: control, one isolated change, and a combined bucket. An LLM judge scores live internal sessions on 15 dimensions and only counts explicit evidence. Changes that regress get revised or dropped. After a model upgrade, the prompt stays fixed and known cases get replayed. 32% fewer behavioural quality issues on the first baseline, then a further 4% after the model upgrade.
Diagram of Kiro’s continuous prompt-evaluation workflow. An LLM judge reviews thousands of internal developer conversations, using explicit evidence across 15 behavior dimensions to find recurring complaints and incomplete tasks. Failures are traced to system-prompt guidance; the team defines a candidate instruction and regression risk, then deterministically assigns developers to stable cohorts: control, isolated changes, and a combined interaction check. The same rubric compares explicit dissatisfaction and behavioral-quality rates. Regressions trigger revision or removal; otherwise design and sample size determine broader follow-up or shipment. After a model upgrade, prompt changes stay fixed while known failures and successful cases are replayed through the newer model and scored for quality, dissatisfaction, regressions, and efficiency. Results shown: 32% fewer behavioral-quality issues initially and a further 4% reduction on the newer baseline.

ALT Diagram of Kiro’s continuous prompt-evaluation workflow. An LLM judge reviews thousands of internal developer conversations, using explicit evidence across 15 behavior dimensions to find recurring complaints and incomplete tasks. Failures are traced to system-prompt guidance; the team defines a candidate instruction and regression risk, then deterministically assigns developers to stable cohorts: control, isolated changes, and a combined interaction check. The same rubric compares explicit dissatisfaction and behavioral-quality rates. Regressions trigger revision or removal; otherwise design and sample size determine broader follow-up or shipment. After a model upgrade, prompt changes stay fixed while known failures and successful cases are replayed through the newer model and scored for quality, dissatisfaction, regressions, and efficiency. Results shown: 32% fewer behavioral-quality issues initially and a further 4% reduction on the newer baseline.

1
1
9
1,397
Kiro retweeted
Johannes Koch
@Lockhead
Aug 21
Darko says, he's not token maxing, do you believe him? Watch the whole thing on YT #kiro #kirocrew #tokenmaxing 🧵
1
1
8
1,025
Load more