pelavo (@pelavo7) header
pelavo (@pelavo7) profile picture

Subagent happiness manager · Product, Code & Infra · Web3 Council @labamille

Joined April 2022
Photo by pelavo (@pelavo7)Photo by pelavo (@pelavo7)Photo by pelavo (@pelavo7)Photo by pelavo (@pelavo7)Photo by pelavo (@pelavo7)Photo by pelavo (@pelavo7)Photo by pelavo (@pelavo7)Photo by pelavo (@pelavo7)Photo by pelavo (@pelavo7)Photo by pelavo (@pelavo7)
The sort of thing your agent writes tests for
The chances of being killed by a landmine delivered by a bear are low, but they are never zero.
76
834
36
14,203
323,701
-----BEGIN RSA PRIVATE KEY----- 4 -----END RSA PRIVATE KEY-----
245
3,997
330
42,311
1,102,862
pelavo retweeted
22
48
1
1,377
28,476
Might as well give voice to it! Thanks to @Toro_Yuma for the idea!
50
877
36
10,270
417,627
Take a look at your favourite skill. Go on, take a look. Check for lines like: - "Make the commit message very detailed" - "Be thorough" - "Make the implementation easy to read" What do these lines have in common? They're no-ops. They do nothing to change the agent's behavior. Agents always write good commit messages. They always try to be thorough. They always try to create easy-to-read implementations. Try removing them. Does the output change? No? Then the line was a no-op. Agent-authored skills are LITTERED with no-ops. Entire paragraphs and sections of text that does nothing to change the agents' behavior. No-ops make skills harder to evaluate, harder to maintain, and burn tokens uselessly. Learn to spot them, test them, and remove them. Your skills will thank you.
189
151
48
3,192
309,213
My current experience with coding models.
74
352
41
9,746
389,216
can I buy this in the metaverse? @facebook
3,163
3,119
1,178
39,984
3,750,773
Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.
7,867
10,902
7,550
148,798
28,092,496
pelavo retweeted
Overthinking
60
284
16
5,638
1,165,719
New art project. Train and inference GPT in 243 lines of pure, dependency-free Python. This is the *full* algorithmic content of what is needed. Everything else is just for efficiency. I cannot simplify this any further. gist.github.com/karpathy/862…
636
3,073
574
24,733
5,236,170
it has started…
Recursive Language Models (RLMs) let agents manage 10M+ tokens by delegating tasks recursively. This Google Cloud Community Article explains why ADK was the perfect choice for re-implementing the original RLM codebase in a more enterprise-ready format →goo.gle/4kjT12E
23
74
2
2,399
368,171
I just built a *functioning* React app by describing what I wanted to GPT-3. I'm still in awe.
441
4,050
1,963
21,828
Here's a 1 GB memory reduction for very long Claude Code sessions Before: `() => controller.abort()` Fix: `controller.abort.bind(controller)`
131
214
91
4,512
1,545,932
pelavo retweeted
The solution to anxiety is action.
1,155
4,916
446
39,145
1,254,843
pelavo retweeted
Setting up Next.js + shadcn/ui + Drizzle + Better Auth takes 3-4 hours every single project. Same lint rules. Same Claude Code settings. Same pages. Same auth flow. Copy. Paste. Debug. Repeat. We all hate it. Sound familiar? 🧵
4
8
1
10
265,651
you will regret not using a monorepo.
137
106
27
2,458
189,378
Replying to @VibeMarketer_
“You should try inviting AI to help you in everything you do… familiarizing yourself with AI’s capabilities allows you to better understand how it can assist you—or threaten you and your job” ~Ethan Mollick Book Summary: Co-Intelligence
1
12
44
5,723
pelavo retweeted
In your 20s and 30s, you will be tempted to prove that you’re doing great. It’s important that you resist performing for an audience that doesn’t care.
163
7,404
176
58,086
901,757