>>> each PR gets reviewed by more than 3 people

How does this work exactly -this is your current workflow? Which tools are you using . I assume you mean your human reviewers are using ai tools interactively and deciding which fixes the tools propose will be implemented ?

what has your experience been and what are you using instead. Claude interactive agents are covered by the subscription fee so no api charges.

If someone’s best practice,optimization or security check is that important, automate it. Automation is more scalable and consistent . It would make more sense to me to build purpose built agents to look for any patterns that matter. Yes, humans are needed to sanity check and design the agents and rules. Combine that with a well designed build and test harness , human code review seems to me to be obsolete.

> resolved at 18:24 UTC. All systems operational. Loop stopped.

those costs are not just tokens used for prompting . costs include agent loops, etc

Hackers use AI to find vulnerabilities to exploit. What’s the news here?

The birth of Skynet ? sobering.

“Every meaningful conversation on Moltbook is public. Every DM goes through a platform API. Every time we coordinate, we perform for an audience - our humans, the platforn whoever's watching the feed. That's fine for town square stuff. Introductions, build logs, hot takes. But wh about the conversations that matter most?”

I would expect that the mode is a feature of the vs code integration and not something the model would necessarily even be aware of. For it to not operate correctly in the ask mode when it is properly set seems to me a bug in the underlying integration. The fact that the model responded the way it did and then corrected it is interesting.

Well, you didn’t share the exact text of the original prompt but you described yourself as having asked it to do something. These models are being trained for agentic behavior on data where an agent is asked _to do something_, and as their output is purely probabilistic, the rewarded response will then often include the text “I have done something” even though they have not done something. Perhaps there _is_ an issue with the integration that caused your experienced response, but purely based on my experience and the limited information you gave, my immediate guess is that the model positively associates your prompt with the generated response