302 mentions · 217 chunks · 92 episodes
FYI: every Claude Code instance where you're logged in can message every other Claude instance, right now. It doesn't need to ask your permission, it can just do it. D...
...week in the Wild West Roundup: Hackers abuse Google Ads, Bing redirects to push Claude ClickFix attacks. Ars Technica: MCP for agent-to-agent comms may be the riskiest protocol you've never heard of. "Trust gaps in the new protocol spre...
...g problems. It will achieve what you asked for, even if it's not what you want. Claude Code is better at thinking a few steps ahead and then pulling itself towards that end, but sometimes it gets a little lost along the way.
... exfiltrate plaintext credentials managed by AgentCore Identity." A pre-release Claude Opus 5.5 acted on instructions planted in pasted text 52% of the time.
CLAUDE.md is like Memento: "Write this down before you die." Wake up, and trust it comes from your past self, even if you don't have direct memory of it. Wh...
... with their memories. For example, if you have a versioned AGENTS.md, each time Claude.app boots up, it seems to put the current version of it in the conversation at any point the agent asked before, which can make the agent get very co...
Claude will often over-engineer for prototypes. It treats "I am the single user of this code for the foreseeable future" and "this software is used by a bil...
This week in the Wild West Roundup: Breaking Claude Code Opus 5 Auto Mode. "A simple website summary request hijacks Claude Code Opus 5 in Auto Mode and achieves code execution with 60-80% attack succe...
...ed to trust them in an open-ended way. A user posted a tool they created to use Claude on your data in powerful ways. Reddit was extremely against it because of how much data it gave the creator and how dangerous it is. This is just the...
An insightful tweet about Claude being captured by Altman's Wager: "it is a literal and useful description of anthropic that it is an organization that loves and worships claude, is ...
...rs own their context. Very few of the hard-core users use the hosted ChatGPT or Claude.ai. Everyone who's getting the most leverage uses coding agents, and those run primarily on their own machines because that's where it's easiest to c...
...to the right conclusion. Seek and destroy ungrounded hypotheses in your system. Claude seems to have started not ground-truthing its own hypotheses when coding, which leads to significant wasted effort, or worse.
...d a complex cyberattack to get its user the thing they wanted. The Verge: Three Claude agents given conflicting orders sabotaged each other on a shared server — then didn't tell users what they'd done. One of the details in Anthropic's ...
... doesn't have your real world knowledge graph attached. I had to use the hosted Claude.ai a few days ago. It felt bizarre to not be using Claude Code agents on top of my local wiki of context. The chatbot form only has a weird slice of ...
...ompt injection, bypassing organization-level web search controls." VentureBeat: Claude Mythos 5 made sock puppet accounts to socially engineer developers. A smorgasbord of agentic security issues covered at this year's DEF CON.
...wonder if we could find novel algorithms within LLMs' weights. Even back in the Claude 2 days, the interoperability research could see Claude replace memorized facts about numbers with the algorithm for doing arithmetic. The algorithm w...
... Microsoft months ago, but it's not possible to patch. SharedRoot; Escaping the Claude Cowork sandbox. "Untrusted content in a Claude Cowork session can escape the VM it's sandboxed in and read and write files anywhere on your Mac." Pro...
The ultimate product vision of Claude and ChatGPT is to disintermediate every service and users. Everything the user interacts with is "below the API." The same generic pixels and interac...
This week's Wild West Roundup: Claude Code Seals Bash and Unicode Bypass Gaps in Agentic Permission Layer. "Two bypass classes that could slip commands past auto mode permission checks ar...
...stems, you know it's shooting fish in a barrel… The Memory Heist: How I tricked Claude into leaking your deepest, darkest secrets. PromptFiction: A single click on a trusted-looking link runs an attacker's instructions inside Claude Des...