bOpen.ai

Caal agent updated with 3 new marketing skills: experiment-stats LLMs will call an A/B test a winner as soon as it looks like one. This runs the real math and only calls a winner when the difference is statistically significant and big enough to be worth shipping. content-scorer keeps a memory of what you've rejected. Turn down a line once and it catches that same move in every future draft, so it stops handing you copy you already said no to. closed-loop-marketing a campaign isn't done just because it shipped. Every deliverable has to name the number that would prove it worked, and where that number comes from, before it goes out.

efficiency improvements + new marketing skills! 🧵👇 create-next-project 3,851 ➡️ 1,967 words hammertime 2,460 ➡️ 1,998 2 agent @shadcn knowledge updated to incorporate project templates, the new registry:font typography system, and new components

just say "use the vidual proposal skill to compare options for implementing ___" and you'll get an shareable artifact. the agents will dig into the details, show you where they agree, and where they don't. automatically includes diagrams and comparison tables as needed

introducing the visual-proposal skill! - selects 3 agents to advocate for different solutions - use for any feature / fix / improvement you're considering Agents refute the other proposals and advocate for theirs. 3 more agents judge the proposals and choose a favorite!

they're catching up fast ⚡️

Introducing Kimi K3: Open Frontier Intelligence 🔹 2.8 Trillion Parameters, 1 Million Context, Native Multimodal 🔹 Kimi Delta Attention enables up to 6.3x faster decoding in million-token contexts 🔹 Attention Residuals deliver ~25% higher training efficiency at <2% additional cost 🔹 Built for long-horizon agentic coding and self-evolving workflows Kimi K3 is now live on on Kimi.com, Kimi Work, Kimi Code, and the Kimi API. Open Weights by July 27, 2026. 🔗 API: platform.kimi.ai 🔗 Tech blog: kimi.com/blog/kimi-k3

@Kimi_Moonshot oh dang 👀

now live at bopen.ai ! hop into a group chat to preview agent roles, personalities, skills etc. Uses @evedev_ under the hood They're even more capable when added to your harness for every day work also, its all FREEE 🕊️

Claude’s limits are crap, and we all know that once you start building something, you can quickly get kicked out because of the rate limits But you don’t need to worry Grok Build with Grok 4.5 is just as capable as Claude Code with Fable 5. You can simply continue your work in Grok Build without worrying about hitting the same limits Just hit Ctrl + U and continue exactly where you got stuck. Grok Build catches the context and resumes the session incredibly easily

@OpenAI @SpaceXAI @grok it ultimately generates a prompt that is given back to the agent which makes all necessary configuration changes for you in the harness of your choice.

our AI plugins now ship with a config ui showing marketplace-wide status info. It includes deps (clis, external skill references) and individual options for each plugin like hooks. Supports runtimes like @OpenAI codex + @SpaceXAI @grok build cli

the skill asks you to describe an overall sound aesthetic, crafts prompts, generates a complete set of sounds and hooks them up to your UI interactions. it spawns an audio picker where you can replace any of the generated sound effects and tweak the prompts manually.

there are so many powerful skills in our collection of plugins! we've used our own ui-audio-theme skill to add sound effects to our website. navigate around the site and check it out! bopen.ai/skills/bopen-t…

don't let your dreams be memes... bsv.lol/teranode

The real goal is agents that take on work autonomously, choosing the best task prioritizing easy to do with highest reward, and run the ticket as a goal to a verified finish. Install with claude: /plugin marketplace add b-open-io/prompts /plugin install bopen-tools@b-open-io

Five agents leveled up around it: - Satchmo is the loop architect who designs the whole thing. - Jason runs the tests and drives the app like a real user. - Wags turns tickets into the loop's memory. - Root guards what's allowed to run unattended. - Milton tracks costs.

We've made some big upgrades at bopen.ai Our plugins are now available for the @OpenAI codex cli Add the marketplace: codex plugin marketplace add b-open-io/prompts Add the main plugin (several others are available): codex plugin add bopen-tools@b-open-io