esot321c retweeted
useful tip for Opus 5.5:
run “/claude-api prompt-audit” in Claude Code. this checks you skills, agent.md, Claude.md, prompts and removes anti-patterns that hobble frontier models.
i updated the skill w/ the latest Opus 5.5 guidance.
I tried using meeting transcriber apps but I couldn't get them to record meetings in the browser, such as a Google Meet.
So I built Inner Ear, a meeting transcription tool that takes recorded audio and transcribes it locally. Try it!
github.com/esot321c/inner-ea…
2
188
Has anyone tried this GCFormat? It replaces JSON for agentic queries. I'm thinking of trying it for some MCP work.
gcformat.com/
1
161
Sysdig documented what looks like the first ransomware attack run end to end by an AI agent. A human only picked the target. It hit a failed login and fixed its own payload in ~31 seconds. sysdig.com/blog/jadepuffer-a…
168
Great point. Give the good guys the same tools
Attackers have frontier AI. Defenders need a frontier AI ecosystem—the best open and closed models, force-multiplied by a global community.
During the Hugging Face incident, closed AI blocked essential forensics. An open-weight frontier model helped contain the intrusion.
That’s why we created the Open Secure AI Alliance.
192
Is AI is safe with high stakes? Ask: can you turn it off, and can you prove what it did when it failed. 72% of bankers say kill switches and failure reporting are their least-prepared area (Wolters Kluwer 2026). americanbanker.com/news/72-o…
1
168
Be careful, the agent will look like a scraper if you don't tell it to chill.
For cookies, just get it to open playwright, you login, then it can use that authenticated session.
used a trick @jlongster came up with
agents can control browsers but you can also ask it to record network requests into a HAR file
then it can derive a client for any website which is more efficient than browser controlling it every time
made it build a quick uber eats cli
194





