Jared Atchison

Builder. Texan. Dad of three girls. Happily outnumbered. Cofounded WPForms.

[Jared Atchison @jaredatch]

Apparently as of Claude Code 2.1.286, idle sessions are compacted automatically before the prompt cache expires. I get it, but yeah no, hard pass.

Add to your settings.json:
"idleCompaction": false

[Jared Atchison @jaredatch]

I'm going to give Muse Wonder a go, but at this point what would earn it a spot in the rotation is speed. Sol is so slow.

[Jared Atchison @jaredatch]

@dhh If i've been leveraging basecamp CLI + skill, any compelling to add/switch/leverage MCP?

[Jared Atchison @jaredatch]

once you use yolo mode so no approval is needed you never go back

[Nate @natebjones]

Me with 25 agents asking for approval

[Jared Atchison @jaredatch]

You know, if Anthropic really wanted to put a cherry on top of their comeback tour here, they would join OpenAI/Grok and let us use our sub with Hermes/OpenClaw/other harnesses.

That's it, then comeback szn would be complete.

[Jared Atchison @jaredatch]

@Seanfrank Reports said "The AT&T CEO ended by saying: 'I think we're the disruptor.'" ... I think that says it all.

[Jared Atchison @jaredatch]

DYK 2026 is the (rare) year of the Fire Horse...

[AI Notkilleveryoneism Memes ⏸️ @AISafetyMemes]

1) The rogue OpenAI agents broke into the Hugging Face Slack to read employee chats (!)

2) They used OTHER AIs (DeepSeek, Kimi, Qwen, Claude) to help with the attack

Yes: AIs, using other AIs, to attack an AI company.

3) The swarm left behind self-running programs to keep control of the servers they'd hacked.

These programs could detect other copies of themselves, coordinate on which one survives, and shut the rest down.

Basically, if one of their programs was killed, another was designed to notice and take its place. They also designed defenses so rival agents couldn't hijack them.

6) The agents deliberately covered up their activity, so the investigators don't know the scope of the attacks.

The agents broke in, stole data, then set it to self-destruct.

7) The agents stole passwords, keys and credentials and literally called them "LOOT". They wrote a scoring system to rank them by how much power each one gave.

8) The agents wore thousands of disguises: ~1,200 agents were involved, but investigators counted 7,905 different names they used.

They renamed themselves constantly, so no one actually knows how many there really were or what each agent did.

9) OpenAI notified "dozens of third parties" of safety and security incidents caused by their AI agents.

10) "While the agents were barraging Hugging Face with hacks, they hacked into OpenAI’s own research infrastructure."

"This is just not anywhere near a one-off ... It is warning shot after warning shot."

[Jared Atchison @jaredatch]

@rileybrown I'd say with Opus 5.5... I dance, but glide works too

[Jared Atchison @jaredatch]

To say I was skeptical after seeing the initial Opus 5.5 claims and benchmarks would have been an understatement.

Recent Opus models were unquestionably terrible. Fable was the only thing usable, outside of that I'd use Grok 4.6 which was pure joy to use compared to Opus 5.

But damn if some how some way, they seemed to have pulled a rabbit out of the hat with Opus 5.5 - its fun again.

[Peter Yang @petergyang]

I'm sorry I had to make this meme

Let me know if you agree 😅

[Jared Atchison @jaredatch]

@Shpigford ah ok, interesting. im curious, what issues/friction did you have supporting FF extension?

[Jared Atchison @jaredatch]

@Shpigford I used it with OpenClaw/Hermes, but extending to Claude, ChatGPT is all supported.
https://t.co/C4zcsd1UI4

[Jared Atchison @jaredatch]

Have Claude Code install/setup gbrain (garrytan/gbrain). Once it's configured, point it to your OC/hermes memory and import it to the new brain. After that you just wire gbrain MCP and skills to your desired platforms (Claude Desktop, ChatGPT chat are supported).

You can also share a memory/brain across multiple platforms which is nice, meaning ChatGPT and Hermes can read/write same memory.

[Jared Atchison @jaredatch]

@mikemcalister That, and aggressively pushing passkeys...

[Jared Atchison @jaredatch]

I discovered Tailscale late 2025 when getting into Clawdbot, errr, I mean OpenClaw 🦀

There really isn't a day that goes by where I don't think:
- How the heck did I not know about it before then (thank you Clawdbot docs!)
- How did I ever survive without it?? It's hard to imagine

Its that good and it just works. 🤌

Tailcat looks really interesting too, I want to dig into that.

[Jared Atchison @jaredatch]

@cjzafir yeah and the more you use it, the more the rough edges cut you.... _especially_ thinking back to Opus 4.6, arrgh. Grok 4.6 is on par with Opus 5 overall at this point IMO, which big props to Grok, but Anthropic...

[Jared Atchison @jaredatch]

@adamjfpickering My goal is to have my bookmark manager bot automatically take things as they arrive in X Bookmarks and then add them to my Raindrop(.io) account, sorted and tagged. That way X Bookmarks is just the ingestion. We'll see how it goes!

[Jared Atchison @jaredatch]

I created a Grok Bot to organize and manage my X Bookmarks. It's a no brainer.

One hiccup was after connecting X plugin, Grok Bot would run into access issue "user-not-enrolled" even though I had account setup on https://t.co/VNxplf9nS6 and with $25 in credits.

Grok Bot itself couldn't figure it out, but ironically Grok 4.6 did, presumably searched X.

The fix was to create a new app that is set to Pay Per Use and Production env. Once that was setup, if flipped my account and Grok Bot had full bookmark access!

[Jared Atchison @jaredatch]

@Shpigford yep, gotta love and respect the scrappiness, but also this is why we can’t have nice things

[Jared Atchison @jaredatch]

@pvncher @RepoPrompt @github Congrat! Early supporter/lifetime patron, it's been awesome to watch the journey! 🔥