Joined March 2025
961 Photos and videos
Pinned Tweet

1
10
63
31,112
Jun 13
🚨 Breaking: Anthropic just disabled their Fable and Mythos models! Fable 5 was available in Kilo Code upon release. Our users and internal team were certainly impressed - but for those who got to try it, did you think it was this powerful? Thread below 👇
8
4
47
3,494
Jun 13
Fable 5 is a remarkably powerful model - but is it powerful enough to warrant this kind of concern? Maybe, maybe not...let us know your thoughts!
1
3
557
Jun 13
Most AI review tools run the same generic rubric on every repo, and miss the things your team actually cares about. Code Reviews now adapt: drop a REVIEWS.md in your repo and the agent enforces your conventions, not someone's default. <<< Live now. >>>
2
1
12
969
Jun 13
The better part? You don't have to write those standards down. Toggle on Code Review Memory and it watches how you respond to PRs. Dismiss line-length nitpicks, act on error-handling comments, and it proposes updating your REVIEWS.md to match. It even opens the PR.
1
4
764
Jun 13
And locally, Code mode now reviews your uncommitted changes before you push. All three are live: blog.kilo.ai/p/code-reviews-…
5
662
Jun 12
Your backlog is optional now. An idea raised on a Tuesday call can be shipped, reviewed, and merged before Wednesday standup. Scott wrote about what that breaks, and what it demands, for @Forbes.
Our rule at @kilocode: if you prompt the agent, you're the author. You review it before anyone else does. Prompting something into existence and hoping isn't ownership. forbes.com/councils/forbeste…
1
900
Kilo retweeted
Join FriendliAI, @nvidia, @MiniMax_AI & @kilocode on July 1st at the FriendliAI office for an evening on the full AI coding stack — GPUs, inference, reasoning models, and agents in production. For founders, eng leaders & builders shipping AI products. Doors at 6 PM, talks 6:30–8, then drinks & networking. Only 80 spots — RSVP to lock yours in! Find it here: luma.com/y454srtx luma.com/y454srtx
3
6
658
Jun 12
Two new coding models and a sold-out token plan this week in Kilo. Kimi K2.7 Code from Moonshot, Claude Fable 5 topping our coding benchmarks, and the MiniMax token plans selling out fast enough to need a new batch already.
3
1
47
2,077
Jun 12
SOTA models should learn to KISS. We planted a cache invalidation bug in a small Next.js app and handed it to Claude Fable 5. It found the bug fast. The fix was valid. The instinct behind the fix is the part worth talking about. 🧵
3
16
1,284
Jun 12
And this pattern keeps showing up. Researchers have documented it, engineers keep posting about it, and Anthropic's own prompting docs note that recent Claude models tend to add abstractions nobody asked for. One theory: benchmarks reward complex tasks, complex tasks need patterns, and simple apps pay the tax.
1
2
463
Jun 12
The Kilo CLI now has a face. 👀 Kilo Console is a local, browser-based UI for managing your projects, worktrees, sessions, and settings. No more hand-editing JSON. <<< Now in beta. >>>
4
2
31
2,247
Jun 12
Open a project, see its git worktrees, launch a CLI session in any of them. Settings show where each value comes from, global or per-project, so you stop guessing which config is winning.
1
5
706
Jun 12
"Another config tool?" No, the opposite. It's the one place your config stops being scattered JSON. Beta's live, and feedback genuinely shapes it: blog.kilo.ai/p/kilo-console-…
4
590