klappy

Thinker, strategist and builder of systems that build systems

[klappy @klappy]

@321K I’ve been using it for over a month and then it disappeared yesterday. What happened? Is this Microsoft’s plan to force Copilot use on GitHub?

[klappy @klappy]

@huskistaken Every chat app since 1988 got this right.
It’s time for AI to catch up.
Full essay + the open-source skill you can add to Claude right now:
https://t.co/gIns6KREUs
Skill:
https://t.co/iBRzsoGPlP

[klappy @klappy]

Sam Altman watched ChatGPT fabricate a mile run time on camera.
“That’s a known issue.”
How long to fix? “Maybe another year.”
I did it in two hours.
Thread: We Forgot to Give AI a Clock 🧵

[klappy @klappy]

If timestamps were in the message schema, the model would already know what time it is. How long you’ve been gone. Whether you’re prepping for a meeting or debriefing after one.
It could even time a mile run — just subtract “stop” from “start.” Platform clock precision. Zero tool calls.

[klappy @klappy]

The full story, the PRs, and the open-source skill:

📝 Article: https://t.co/3T5UYjmDj4

🔧 Skill: https://t.co/8B6QtZqfkd

🔀 Agent-authored PR: https://t.co/9QYB2Sa3FF
What started as “does this work?” became a production bug fix, 3 governance docs, and a reusable skill — in one session.

[klappy @klappy]

Anthropic launched Claude Managed Agents yesterday. By this morning I had an autonomous agent running my epistemic governance system on their infrastructure.

60 minutes from “what is this” to a working prototype that proved value.

But the speed isn’t the interesting part. 🧵

[klappy @klappy]

The whole workflow — spinning up agents, sending tasks, parsing results — was trapped in one session’s context. Next conversation, it’d all be gone.

So I packaged it into a reusable Claude skill. Now any session can dispatch agents without rediscovering the API.