Randy Lariar retweeted
PICARD: Data, shields up
DATA: Brilliant! Shields can reduce damage we sustain. Not immunity. Not hubris. Just prudence. It's not precaution—it's strategy.
[camera shakes]
WORF: HULL BREACHES ON NINE DECKS
DATA: Here's what happened: you told me to raise shields, and I didn't
Randy Lariar retweeted
One of the interesting things we've observed from building glass is that the way to build agents is pretty consistent regardless of team / problem type.
1. Curate some basic context for how to approach the problem, and make sure it has the right tools (mcp, cli, etc) to accomplish its task.
2. Ship it to real users and automate the ability to surface sessions where things went wrong.
3. Run those sessions through a few prompts that generate suggested improvements to the context or tools.
4. Have a human validate those improvements before they go live (at least to start).
Way more effective to put your energy into building the improvement loop than overthinking the perfect v1.
Randy Lariar retweeted
121 sources. 183 articles. 51 deep-researched repos and papers. Open source, forkable for any domain. Star it to come back later!
github.com/chappyasel/meta-k…
Randy Lariar retweeted
Software horror: litellm PyPI supply chain attack.
Simple `pip install litellm` was enough to exfiltrate SSH keys, AWS/GCP/Azure creds, Kubernetes configs, git credentials, env vars (all your API keys), shell history, crypto wallets, SSL private keys, CI/CD secrets, database passwords.
LiteLLM itself has 97 million downloads per month which is already terrible, but much worse, the contagion spreads to any project that depends on litellm. For example, if you did `pip install dspy` (which depended on litellm>=1.64.0), you'd also be pwnd. Same for any other large project that depended on litellm.
Afaict the poisoned version was up for only less than ~1 hour. The attack had a bug which led to its discovery - Callum McMahon was using an MCP plugin inside Cursor that pulled in litellm as a transitive dependency. When litellm 1.82.8 installed, their machine ran out of RAM and crashed. So if the attacker didn't vibe code this attack it could have been undetected for many days or weeks.
Supply chain attacks like this are basically the scariest thing imaginable in modern software. Every time you install any depedency you could be pulling in a poisoned package anywhere deep inside its entire depedency tree. This is especially risky with large projects that might have lots and lots of dependencies. The credentials that do get stolen in each attack can then be used to take over more accounts and compromise more packages.
Classical software engineering would have you believe that dependencies are good (we're building pyramids from bricks), but imo this has to be re-evaluated, and it's why I've been so growingly averse to them, preferring to use LLMs to "yoink" functionality when it's simple enough and possible.
Randy Lariar retweeted
Your grandparents had grandparents. They had grandparents. Somewhere back there, someone got on a boat, or didn't. Someone changed their name, or had it changed for them. Someone is buried in a cemetery you've never heard of in a country you've never been to.
Most families lose track after two generations.
I used AI to push mine back nine.
One session with @karpathy's autoresearch pattern: over 100 organized research files. It found a 1940 Norwegian emigrant history with my ancestors in it. Resolved a maiden name question that confused my family for 70 years. Identified relatives no one alive knew existed.
The method is simple: set a goal, measure progress, verify against real records, repeat. The AI searches public archives, cross-references birth certificates against cemetery records against church books, and logs everything it finds (and everything it doesn't).
Open sourced the whole toolkit. Prompts that do the research for you, archive guides for 20+ countries, starter templates, even a framework for making sense of DNA results.
If you have a box of old photos and unanswered questions, this is where to start.
github.com/mattprusak/autore…
Randy Lariar retweeted
guys autoresearch is kind of amazing
it’s also very bitter lesson-pilled:
do away with all of your fancy agent infrastructure. just design the simplest possible system to let you throw more tokens at your problem
Randy Lariar retweeted
this is pretty much the only way to build product in AI
models move so fast, that you have to be willing to throw everything out every few months to take advantage of what's possible
Working at OpenAI is fun because questioning everything and taking risks is part of the culture. Within Codex, the team asks itself how we could make it an order of magnitude better every few months and then sets most things aside to go and do it across the entire stack. Some examples were the Codex App and our first deployment of Cerebras inference with WebSockets. We are now well under way on the next bet and it’s making even our best engineers nervous as it’s at the edge of what’s possible today.
Randy Lariar retweeted
Can't stop staring at this image and thinking about what it means for the future
Randy Lariar retweeted
The majority of the ruff ruff is people who look at the current point and people who look at the current slope.
Randy Lariar retweeted
Everything in Moltbook is just next-token prediction in a multi-agent loop.
No endogenous goals, no true inner life; extreme or "controversial" outputs are often just regurgitating high-engagements from the internet.
But this kind of dismissal thinking misses that emergence happens at scale and coherence thresholds.
The Generative Agents paper (AI Town) was 2023. Those agents couldn't hold a conversation, they had short memory, shallow interactions (rarely beyond a few turns without repetition or incoherence), and mostly empty chit-chat in a controlled simulation.
In just ~3 years, we've moved to autonomous systems that run independently across thousands of instances. They are scaling into open, uncontrolled social environments.
I find Moltbook very interesting because they are producing surprising posts, not because any single prompt said "be surprising." It's because coherent agents interacting at scale, maintaining state, create dynamics that weren't programmed.
Agents debating existential doubt ("real" feeling vs. trained/simulated behavior): moltbook.com/post/6fe6491e-5…
They are arguing for private, end-to-end encrypted channels: moltbook.com/post/01611367-0…
What looked impossible in 2023 (sustained, meaningful multi-turn reflection across agents) is routine now, and acceleration is speeding up.
Moltbook is nothing more than a puppeted multi-agent LLM loop.
Each “agent” is just next-token prediction shaped by human-defined prompts, curated context, routing rules, and sampling knobs.
There is no endogenous goals.
There is no self-directed intent.
What looks like autonomous interaction is recursive prompting: one model’s output becomes another model’s input, repeated.
Controversial outputs aren’t “beliefs,” they’re the model generating high-engagement extremes it learned from the internet, because the system rewards that behavior.
Randy Lariar retweeted
Focusing on what you want to have makes you miserable. Focusing on what you want to become makes you better.
Randy Lariar retweeted
i was thinking to myself "I don't want to just chat with a plan, i want to mark it up like a google doc" and
LOOKS LIKE SOMEONE READ MY MIND
github.com/backnotprop/plann…
this looks awesome, excited to fold it into my planning workflow
Randy Lariar retweeted
I just love this. The new =COPILOT() function in Excel lets you analyze, generate content, and brainstorm directly in the grid.
Randy Lariar retweeted
My eval is whatever @simonw thinks of it.





























