AI Digest
Source Posts
N
Agent is on another level
In a few minutes, this run was able to output 240,000 data points
It would have taken a whole team multiple weeks to gather this much information https://t.co/Kb37Wuq4GK https://t.co/KDy9vr90Oy
K
Opus 4.5's secret superpower: Playwright testing.
It navigates pages, captures snapshots, checks console errors, tests interactions, and knows when to pause for your input.
Added this to my code review workflow. Now reviews actually verify the UI works.
https://t.co/5Wa2zV7HzF
A
I've been telling anyone that will listen about how amazing Agent Skills are.
Today, we're making skills an open standard that any AI tool can use to make repeatable workflows better and more powerful. https://t.co/iQDApEpJG6
O
@alexalbert__ Cool, we are collecting the commonly used ones and resources since Skills launched. https://t.co/XW7Tsq51bU
J
I’m surprised more people aren’t trying cass. It’s an incredible timesaver for people and agents if you use a lot of coding agents. Works out of the box with every coding CLI, auto detects and ingests/indexes everything in seconds, instant search-as-you-type results, robot mode… https://t.co/4mVjxjCu9t
T
Curated list of resources for customizing Claude AI workflows
https://t.co/iNltClhIa3 https://t.co/X5erABaxdP
🎭
Mans put two Claude on a server and letting them rip. Update your threat model accordingly https://t.co/K1fHLnFYEx
J
So... Postgres is now basically a search engine?
pg_textsearch was just open sourced. It enables BM25 to search your database.... massive upgrade for key word search.
Google uses BM25 in their search engine.
Claude told me: "if you're already on Postgres, you can now skip…
K
sleep time. codex takes over. overnight nightly run that reviews the whole codebase in batches, split across parallel tmux sessions. https://t.co/D1XwrdaoxH
R
Kernel CLI has native computer-use commands (i.e. no Playwright).
It’s intuitive enough that @opencode + Opus 4.5 can just drive a browser and read the HN front page.
(cc @RhysSullivan) https://t.co/ZhcYxGt1b7
K
This video IS the reason why apps succeed. Connor’s back at it again spilling all the secrets 😉
He even talks about our app RidePal, and how we’ve gotten over 100k users so quickly.
We use the same technique he uses to make over 100k MRR. Let me know in the comments what you… https://t.co/FRHiY5PNw6
N
This looks INSANE!
Agent native beads cli just got a major upgrade
Check out @doodlestein new beads viewer
Omg looks beautiful https://t.co/jXbHhXkQyt
D
Can’t stop playing around with this effect.
It will be a game changer for landing pages 😳 https://t.co/Af9z4WtRQH
E
The future is going to be AMAZING with AI and robots enabling sustainable ABUNDANCE for all!
R
@tannerlinsley This is the only one that’s stuck for me, just renting a VPS and sshing into it works well it’s a surprisingly good experience, there’s a guide in the replies
https://t.co/xZYU3UVHaB
J
Be sure to update the blurb in your AGENTS dot md file (although the tool now automatically detects this and asks if you want it to add it for you!). The new blurb is longer but absolutely worth the extra context window tokens:
### Using bv as an AI sidecar
bv is a graph-aware…
0
https://t.co/5tkhh7vDFU
It's live. it will be significantly improved, cleaned up etc.. by tomorrow, will put on npm soon but want to get it good first.
Orchestrator patterns help run models over very long periods of time, save costs, and increase what you can put out.
They… https://t.co/ikmlfQfwFB https://t.co/ZzC3P1Ex6s
K
The amazing @danshipper built a full iOS reading companion in a week. Didn't write a single line of code. Didn't even look at the code.
The secret: features aren't functions anymore—they're prompts. You define outcomes, the agent figures out how.
We open-sourced a skill that…
R
this is Opus 4.5 controlling my browser, running and debugging my codebase within lovable.
right now, Opus is essentially working with another Opus on the app, since lovable agent is *also* Opus 4.5. hahaha.
i think we're cooked, chat. https://t.co/od8EtPOw8n https://t.co/l0JVkZGV6P