data analyst learning web dev / plant-based / movie enjoyer
Joined December 2021
- Tweets 1,939
- Following 572
- Followers 522
- Likes 6,134
265 Photos and videos
Pinned Tweet
May 4
finally completed my first homelab build. 10" mini rack. she so prettyy
2
1
7
516
rip radio raheem. good to be reminded how powerful the mob is, for both love and hate, kinda just depending on the day. also a universal truth that ppl go crazy when it's hot out.
I think we're all entering a very new york time in our lives
doing my part by watching Do the Right Thing today (doubly so 'cause today's gonna be a scorcher)
22
I think we're all entering a very new york time in our lives
doing my part by watching Do the Right Thing today (doubly so 'cause today's gonna be a scorcher)
52
oscar gabriel retweeted
It’s funny, motocross at the White House is, like, fine with me. But here’s an example of something that is genuinely unforgivable but that a lot of people seem to have moved right past.
80
711
6,085
168,110
Jun 13
if I had a nickel for every time anthropic suddenly competely gutted my ability to use their latest and greatest model in the evening before a weekend, mid-generation...
I'd have two nickels
Jan 9
uhh did anthropic just completely kill using your claude subscription w/ opencode?? this was mid-generation
1
127
Jun 13
uhhhhhhhhhh
Jun 13
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.
Access to all other Claude models is not affected.
We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.
Read our full statement: anthropic.com/news/fable-myt…
24
Jun 12
not convinced anyone's having more fun vibe coding than hank
54
Jun 12
lately, I've been turning better-chat, which was a sorta basic chat app I made last year (per user DOs and bundled MCPs were cool ideas!), into better-agent, which is neither a "chat app" nor a hyperactive "agent orchestrator"
it's intended for focused thought work. each "thinkspace" is one @CloudflareDev Think agent seeded with one bounded goal, with one bespoke set of tools and permissions. it lives in the web, and works independently while you're away until its goal is achieved. then you move on to the next thing, responsibly managing your human focus along the way.
6 Oct 2025
alright it's finally here
better chat through durable objects
per-user isolated DOs storing chat thread and message data, for a seriously performant AI chat experience
With coding agents taking over, I found myself using the big web apps less and less b/c they don't have access to the kinds of tools I've gotten accustomed to, so I built a web experience w/ first-class support for @Context7AI, @CloudflareDev docs MCP, @ExaAILabs web search, and more.
and also, who doesn't love durable objects?
1
1
15
7,271
Jun 12
better-agent is how I personally want to imagine the future of agents. rather than gargantuan shotguns that can do anything and automate all the tedium out of your life, thinkspace agents are for when you need a space to sit down and work through a big problem with some help from something that only knows how to help you with that problem.
your attention is your most important finite resource; manage it wisely with better-agent.
1
5
300
Jun 12
still extremely a work in progress and not live yet, but follow along here github.com/oscabriel/better-…
1
5
255
Jun 11
I haven't experience a single prompt refusal from fable 5 when using it via @opencode zen, thx besties
65
Jun 10
what's up with ppl putting the after photo first now
who hurt you
1
37
Jun 10
skills are programs for your agents btw
/replicant is my third or fourth iteration on this idea. I've previously made cli tools, web apps, etc that accomplish the same thing, but turns out you can put all the same logic in a skill and the agent knows what to do with it
it's my version of a repo search skill, with some very light persistent git clone management thrown in
sets itself up on first run, then it knows exactly where to look and what bash commands to run to speedily find and read source code and docs to answer your questions about any open source repo
1
118