279 mentions · 202 chunks · 84 episodes
...tion, chaining them together to escape the sandbox and execute arbitrary code." Claude Code bypasses safety rule if given too many commands. Google Addresses Vertex Security Issues After Researchers Weaponize AI Agents. Ars Technica: He...
...crue to skills or harnesses? Or possibly something else? Agentic harnesses like Claude Code have a few components. 1) the core agentic loop, 2) a ton of code to produce a janky TUI. As the code leak showed, there's no magic in Claude Co...
A friend who conducts multiple agent swarms recently had his Claude Max account turned off by Anthropic. He wasn't doing anything against the Terms of Service… and wasn't even using Claude as aggressively as some peop...
The power of tools like Claude Code comes from the open-endedness of LLMs' reasoning being merged with the open-ended capability of the CLI. That explosive power is combinatorial. ...
This week in the Wild West roundup: Claude Extension Flaw Enabled Zero-Click XSS Prompt Injection via Any Website OpenClaw Agents Can Be Guilt-Tripped Into Self-Sabotage. "In a controlled expe...
The Claude Code Max pricing model implicitly expects usage to stay fixed. That is, to have inelastic demand. However, the demand is highly elastic. The cheaper ...
... catch up. When you have this mania, you get anxious when you're away from your Claude Code sessions on your computer. A friend with decades of experience managing financial portfolios described it as the same feeling. Debilitating, man...
The power of tools like Claude Code mainly arises from the combinatorial power of the CLI. The power of LLMs as a catalyst for unleashing the inherent (but intimidating) combinator...
...West roundup: Claudy Day: an exfiltration that can happen entirely in a default Claude session. A rogue AI led to a serious security incident at Meta. Vulnerability in MS-Agent AI Framework Can Allow Full System Compromise. Snowflake Co...
...ed contempt for its users. It's kind of hilarious especially when juxtaposed to Claude, who is eager to please to an almost absurd degree.
Claude Code distills ephemeral tokens into durable value. This is unlike chatbots. As a user, you pay a one-time cost to get a durable good. Tokens are an e...
The subscription I hold most dear currently is my Claude Max subscription. If someone threatened to take it from me, I wouldn't be responsible for my actions.. But this loyalty is not so much because of Cla...
Anthropic is clearly pushing to own the scaffolding that everyone uses. Claude Code is best with Max, and they'll disable use of the description in other harnesses. Claude Code apparently also works to thwart the KV cache of oth...
Claude Code is great at deobfuscating code. Deobfuscating is an exercise primarily in patience. LLMs have infinite patience. A kind of funny mental image: C...
Claude Code can fuel a new kind of mania. This is unlike the "the chatbot is my friend and a God" mania inspired by ChatGPT 4o. It's a kind of hyper-compete...
...hinking you can't short circuit the process. Agents can help but you can't say 'Claude, build me novel insights.'"
Capturing the power of Claude Code for non-developers will unlock a lot of value. But that will be very hard to do in a naive way. Much of the open-ended power of Claude Code is a...
Overheard this week: "I trust Claude's judgment calls more than the average American's"
My brother-in-law (who has no engineering experience) discovered the joy of Claude Code. "I have always wanted to be able to take my visions and turn them into reality, but I had this knowledge gap that was impossible to bridge shor...
When you use Claude Code, you have all of your data checked in to your own repos. So you can switch models at any point. The model is only important when you make change...