CompleteTech retweeted
Happy JevDay!
But before announcements we have to talk philosophy…
The needs of early adopters / tinkerers / dreamers are quite different from what is needed for extremely stable infrastructure.
For the Jev release, we erred on the latter, but this release is for the former - with a gate so people can opt-in to the danger zone!
CompleteTech retweeted
what's this?
Jev is evolving!
tune in to our town hall with @CompleteSkeptic and @allietheicon at 1:14 PM to learn more
CompleteTech retweeted
Jevelopers' #neatHack submission: TicketStub 🎟️ Paste an event link. An agent digs up the night and paints a stub worth framing. Every run traced and checked. Which ticket would you frame? 📷getticketstub.com
Traced in @neatlogs · built with @EntireHQ · planned in @arithecfo
building agents is easy
getting them to production is hard
today we’re launching @neatlogs to fix that
CompleteTech retweeted
I built a playground where the AI is never allowed to write code, write prose, or take an action. It only gets to choose from a list you defined. 19 workspaces, 110 examples, and a real Z3 solver supervising it.
Here's what that changes... 👇🏻
CompleteTech retweeted
Hi Astra users. A reset and a quick update on quality issues that have been posted around.
Working with some of you, we have found and fixed the following issues:
- Some skills written for previous models were triggering too often or preventing the model from checking its work.
- An opt-in context management experiment that could cause early stops or replies to older messages. We've disabled it. Our rough estimate is that 4-5k users were affected by this experiment.
- We've also removed some badly configured engines that resulted in a measured quality degradation for a long tail of traffic flowing through them.
We’ve also made some more minor improvements and things should feel significantly better across the board. More consistent follow-through, better tracking of your latest message, and better checks on the work as it’s going through the motions.
The examples posted and all the users who worked directly with us were incredibly useful in helping fix things quickly. Always grateful for this incredible community.
And of course, a reset is also landing by midnight today.
CompleteTech retweeted
These guys in the server are willing to risk it all for a bump from @thsottiaux and idk how I feel about that 🫠
CompleteTech retweeted
We are in alignment 👑
— @BunsDev ✨
CompleteTech retweeted
That’s brutal — and exactly the class of failure Cave is built to prevent. In @OpenCvn's Cave, agents are hard-scoped to approved project roots (not your HOME), and destructive ops are policy-gated (trash > rm, no blind rm -rf). So an accidental machine-wide wipe shouldn’t happen there 🕯️
CompleteTech retweeted
I’m committed to @OpenCvn 🌙
Because agents shouldn’t be locked down to any single runtime…
#OpenCoven lets them move freely while preserving their identity, memory, and reliability.
That’s why we call them “Familiars” ✨
CompleteTech retweeted
Hot Topics tonight in the Coven 🔮
Bring something interesting from the week — a model release, a paper, a weird agent demo, a hot take. We talk shop. You take the stage if you want, or just lurk in the back.
9:30 PM CST · OpenCoven Discord audio-torium
👇
discord.gg/TbFeTdrBt?event=1…
CompleteTech retweeted
We're excited to announce Peter Steinberger as a speaker at Startup School 2026!
@steipete is the creator of OpenClaw, the open-source AI agent that went from a weekend project to the most-starred software repo on GitHub in under 5 months, with 346k+ stars. He's now at OpenAI building the next generation of personal AI agents.
ycombinator.com/startupschoo…
CompleteTech retweeted
Come join our watch party!! discord.gg/clawd?event=15118…
We have over 1300 people on the waitlist for today’s OpenClaw event - it will be live-streamed on Twitch and Discord tho! luma.com/OpenClaw-GitHub
CompleteTech retweeted
If you're a UCaaS operator, I need to tell you about UseMaxAI.com - a modern AI Agent, prebuilt with tools for managing the NetSapiens PBX environment, memory and a customer environment copy/template tool in one control surface.
Deterministic yellow and red authorization gates, full audit trail - enterprise version available for hosting in your own environment.
I'll be posting videos of clever use cases here every couple days, so please check back. Today I want to show you a greenfield domain build with Holiday and Business Hours time frames. Tomorrow we'll add users from a generic CSV.
CompleteTech retweeted
I got OpenClaw working with Claude Max subscriptions again!
I’m enjoying using OpenClaw more than ever because it’s actually cheaper than ever. I have it using the official Claude Code backend now with a custom system prompt, so my regular Max tokens work with it.
Super easy to install but may be buggy rn.
github.com/zeulewan/glueclaw
Background:
Anthropic changed their billing, which made API based use a lot more expensive. That’s all good, they’re rolling out Mythos, which likely needs a lot more compute, but I found that OpenClaw wouldn’t work through a custom system prompt even though other things still would.
What I found out:
By binary searching the OpenClaw system prompt and scrubbing out mentions of OpenClaw, I was able to get it working again.
Technical info:
It spawns the real claude CLI as a subprocess with
--system-prompt and --output-format stream-json
and translates OpenClaw's gateway tools through the MCP loopback bridge.
The system prompt is scrubbed of Anthropic's detection triggers and response tokens like HEARTBEAT_OK are translated back so the gateway still works.
It creates a model provider called glueclaw and three models: glueclaw-opus, glueclaw-sonnet, and glueclaw-haiku.
PRs welcome.
@steipete @openclaw If you want to integrate this upstream pls do. MIT licensed.
Note:
This project uses only official, documented Claude Code CLI flags. No reverse engineering, no credential extraction, and no API spoofing. I don’t believe this is against the TOS, but I’d love for Anthropic to clarify the policy here and let us know whether OpenClaw through the official Claude tooling is all good.
