In my rocky training montage currently
Denver ∈ 🇺🇸
Joined April 2016
- Tweets7.8K
- Following356
- Followers1.5K
- Likes47.3K
I’m mid battle with this pc rebuild, and let me tell you:
I was NOT prepared.
I’ve gone through several trials and tribulations, but I see the finish line.
Most people don’t understand how hard it is to be a true wordcel.
To be truly good at something, you have to be immerse yourself in the art. Study it, care about it deeply.
The fine nuance involved escapes most people.
2
143
My case in point exactly, but this is a HUGE blunder for a company already leaking revenue like a sieve.
Don’t get me wrong: I’m a HUGE cursor fan, and love the product.
I’m allowed to point out fouls on the play here though
Cursor AI may be in material breach of contract with their new Composer model, which is generating buzz online for reportedly reaching Opus-level performance. It’s alleged that the new model is a fine-tuned checkpoint of Moonshot’s Kimi K2.5.
If true, the original model is licensed under a modified MIT license containing a clause restricting commercial use by products with over 100 million monthly active users or more than $20 million (or equivalent) in monthly revenue.
The requirement for triggering that clause is simply to prominently credit “Kimi K2.5” in the product or service’s UI.
Cursor could now face serious PR and legal issues simply because they couldn’t be bothered to cite the underlying team’s work. This is open-source 101....
Talk about being hoisted by your own petard.
4
182
It’s just Kimi K2 lmao
🚨 Cursor just dropped Composer 2..
their own AI model.. not Claude.. not GPT.. their own..
and it beats Claude Opus on coding benchmarks.. at a fraction of the cost..
a code editor with 50 people just outperformed a $30 billion AI lab.. at coding.. which is supposed to be their whole thing..
the vibe coding era just got an upgrade..
Beads might change everything about how I develop software agentically.
I've known about beads for a while. It's a critical component of another project I've memed about a while back. (Gas Town)
At it's core though, beads might be the solution to the endless plan md files that litter my all my repos.
As I walk on my treadmill and watch my Cursor regurgitate thousands of lines of code, I feel hopeful.
I'll keep you guys updated.
github.com/steveyegge/beads
1
92
I think that Test Driven Development is almost a necessary requirement to use LLMs for development.
If you need to black box something with AI, you better clearly define the inputs and what you expect the output to be.
Otherwise things get real weird.
3
133
This reminded me of the time one of my close moots found a girl on here a couple months ago and they both quit X…
There’s hope for you out there, anon.
3
209
This is what I’ve been looking for!
Hell yeah
IronClaw (@near_ai, Rust) is the most architecturally serious alternative. Built by @ilblackdragon as a direct response to OpenClaw's security failures. Tools and channels run in isolated WASM containers with capability-based permissions. Credentials live in an encrypted vault and are domain-scoped. That directly blocks the exact exfil vector, where a ClawHub skill was silently curling credentials to an attacker-controlled server. Auth is handled entirely outside the LLM flow. All arbitrary code runs inside Docker containers. Network calls are intercepted and checked for data leakage and prompt injection. Gateway defaults to 127.0.0.1 instead of OpenClaw's catastrophic 0.0.0.0 binding. And it leverages NEAR AI infra for confidential and anonymized inference. Apache-2.0 licensed, no OpenAI ties.
3
162
RIP Openclaw
Peter Steinberger is joining OpenAI to drive the next generation of personal agents. He is a genius with a lot of amazing ideas about the future of very smart agents interacting with each other to do very useful things for people. We expect this will quickly become core to our product offerings.
OpenClaw will live in a foundation as an open source project that OpenAI will continue to support. The future is going to be extremely multi-agent and it's important to us to support open source as part of that.
1
1
79
That is so fucking metal
We love to see it
'ONE PIECE' Live Action's Sanji Actor, Taz Skylar:
"I didn't drink water for almost 2 days. And didn't eat for 3. The oxygen mask is keeping me awake so I can keep pumping it up in between takes! One of the most extreme things I've ever done by far for a scene..."
1
174
People are trying to use linear attention because it’s faster and scales better.
It suffers from context quality issues.
The current battle is testing mixed sparse/linear attention in different ratios and configurations to find a sweet spot between quality and speed.
Linear attention is fast, but you drop context to do so.
Minimax actually found this out with their first linear attention model, which trained well on benchmarks but performed abysmally with reasoning.
On paper? Good.
In practice? Awful.
1
74