Gil
153 posts
ITS ALWAYS THE SIMPLEST FEATURES THAT HIT THE HARDEST.
dynamic resizing coming to CNVS.
your agents deserve a workspace that inspires you to create beautiful projects.
stop vibe coding in the dumpster.
Max Blade@_MaxBlade
if vibe coding was actually a VIBE : yes you can run remote canvasses straight on your vps and ship to production like a psychopath. yes its built in swift and uses very little memory and is insanely performant. yes it has agentic voice control with gpt realtime 2. yes their is bidirectional agent / canvas control via mcp and cli. yes their is built in shared memory system without bloat. made with love by a dad in his basement who is tired of vibe coding being slow, boring, and constrained. CNVS coming soon.
English
so far fable feels to me like the first real non-incremental step change in capability we've seen since opus 4.5 / gpt 5.2, a little over six months ago
anthropic absolutely COOKED with this one
a little sad for openai - i'm sure 5.6 is going to be good, but not this good
English
@vimtor @kitlangton Are they OSS? If so, can you link them? Effort is easier to recognize than describe, and I’m trying to stop being bad at both.
English
i've been working on 2 different codebases that @kitlangton started and they're a such delight
using effect helps, but so do the effort he put into instructions, setup scripts, tests and architecture
good reminder that ai amplifies people who have a high bar for their work
English
Everyone keeps asking “what is your loop” but nobody’s asking “how is your loop”
English
@kcosr @OpenAIDevs i was in the middle of planning something similar when I came across this post haha. thank you for saving me tokens look forward to learning about all your work!
English
Since @OpenAIDevs are leaving us CLI users behind, here is a CLI tool that allows any agent to manage threads, review message history and set and clear goals (even for itself). I use it with my personal assistant to check status and spawn/resume threads via 2-way voice. 😜👇
Guinness Chen@guinnesschen
If you ever get tired of managing your Codex threads, just let Codex manage itself! Codex can now create threads, search them, organize them, pin the important ones, and spin up worktrees for parallel tasks.
English
@dillon_mulroy @beginbot Ruby has a great yaml parser. Once gippty start with one ruby scripts it slowely creep in more
English
@trashh_dev why you never talk about AI? I feel like if you did you post more memes. I miss the old trash🤡
English
right now I do this by structuring as much context in a workspace, setting very clear requirements & contracts upfront, and working with an orchestrator agent to launch workflows/goals/loops that all basically do the same thing - intake the context, begin whatever implementation, adversarial review + reimplementation loop, varying verification throughout etc.
Peter Steinberger 🦞@steipete
Here’s your monthly reminder that you shouldn’t be prompting coding agents anymore. You should be designing loops that prompt your agents.
English
the same grifters who told you to buy a mac mini to run openclaw are quietly selling those machines at a loss now.
everyone who bought a used 3090 instead is sitting on a card worth more than they paid for it. compute that does real work appreciates. hype boxes depreciate. cheers to everyone who went gpu.
English
It’s true, most folks outside our bubble love and use Claude and in term she opus 4.8
In my humble opinion, 5.5 is still the better model… my workflow hasn’t changed. Opus for UI only.
OrcDev@orcdev
I'd say this is one of the realest AI polls it shows how AI is actually being used in the real world, not just inside our little X bubble whether people like it or not, Anthropic is winning where it matters most: daily usage
English
QME
@NYCMayor @ZohranKMamdani @nycEM @NWS @NYCWater @NYC_DOT @NYCSanitation The only flood warning I care about is the one for Seventh Avenue after Banner 3
English
Heavy rain and thunderstorms may hit New York City this afternoon and continue into tonight.
Here's what we're doing to prepare:
@NYCEM is coordinating citywide response efforts and closely monitoring conditions with @NWS.
@NYCWater, @NYC_DOT, and @NYCSanitation are clearing catch basins across the city, with extra attention to flood-prone neighborhoods like Hollis.
@NYC_Buildings has issued a severe weather advisory and is reminding property owners and builders to secure their sites.
@NYPDnews tow trucks are on standby to assist stranded motorists.
@NYCParks forestry crews are ready to respond to reports of downed trees.
@NYCHRO stands ready to help communities recover if needed.
Here's what New Yorkers can do:
Sign up for emergency alerts by texting NOTIFYNYC to 692-692.
Use caution if you're traveling.
If flooding occurs, move to higher ground immediately.
Check on your neighbors, especially seniors and those who may need extra support.
Our city is strongest when we look out for one another. Stay safe, stay informed, and take care of each other.
English
@PaulNeverovsky Nice! If I was to point an agent to learn from the file any tips and insight on how to prompt it? want to use it as a learning experience.
English
never heard back from anthropic after making this video, even though i applied for a product design role on their website.
today I'm making the figma file with the final UI and raw exploration open source.
grab it here before I change my mind
figma.com/community/file…
Paul Never@PaulNeverovsky
Anthropic just leaked a new Claude app design, and it’s crazy good
English
The assault on an NYPD officer outside Madison Square Garden last night was unacceptable.
New Yorkers are rightfully excited about the Knicks' historic Finals run, and we want fans to celebrate this moment together. There is, however, no place for violence, and no tolerance for attacks on police officers.
Thank you to the officers who worked to keep fans safe throughout the night, and we wish the injured officer a speedy recovery.
The overwhelming majority of New Yorkers celebrated responsibly, and I urge everyone to continue doing so as the Finals return to New York.
English
@simpsoka FR: promote session to durable role.
Some threads shouldn’t end. They should get promoted.
A child session may start as a task, then become the person who owns that area. App should let me promote it: keep the context, give it a durable role, and link it back to the coordinator
English
made a new agent skill callled /docs-to-types.
@mattpocockuk's /grill-with-docs is excellent for getting domain language out of your head and into context docs, but those docs still rely on the next agent (running /to-prd or whatever) correctly parsing the intent and ambiguities inherent to prose.
/docs-to-types sits after grilling and before implementation, and attempts to disambiguate as much as possible from the md files and codify it into types before starting on implementation work.
it's my attempt to force clarity out of the context docs and into types, seams, adapters, errors, call stacks, etc, based on the agent workflows I've seen from @dillon_mulroy and others.
as a talk I watched earlier this week from @filipsodic said, "if you know it, the types should know it."
English