Andre Carrera

491 posts

Andre Carrera banner
Andre Carrera

Andre Carrera

@andr_ec_

Follow for AI and Elixir building in public. Dev/Founder of Lambdo.

Utah, USA Katılım Ağustos 2024
410 Takip Edilen110 Takipçiler
Andre Carrera
Andre Carrera@andr_ec_·
@dexhorthy your story of the nightmare of refactoring and writting patterns by hand has been literally the last two weeks of my life. refactoring was easier with Fable, but still a ton of HITL verification reading code etc. and I'm not even done yet
English
1
0
1
2K
Andre Carrera retweetledi
dex
dex@dexhorthy·
gave up on waiting on nikita for the articles fix - part 1 is here part 2 is coming x.com/i/article/2078…
English
82
215
1.8K
529.6K
Mike Hostetler // Chief Agent Officer
Mike Hostetler // Chief Agent Officer@mikehostetler·
Where do you store your Skills? Local to the project, global in .codex? When will we have a package manager for Skills? Right now I'm manually copying skills to each project, curating what I need and what I don't - no global skills - but this feels messy
English
12
1
10
1.4K
Andre Carrera retweetledi
Austen Allred
Austen Allred@Austen·
“We told the AI models to get really, really good at ExploitGym: (a benchmark that measures how well they can execute cyberattacks). You’ll never guess what happened next.”
English
4
2
32
2.5K
Andre Carrera
Andre Carrera@andr_ec_·
@mikehostetler plus it's tons of setup. I got it to a point that it's usable but took a lot of failures to get to that point. I agree a better abstraction would help. nspawn is my current workaround, reflinks everything as well, so super fast and low disk usage
English
0
0
0
9
Andre Carrera
Andre Carrera@andr_ec_·
@benhylak oh no our model is so big and bad and powerful I'm pretty sure you'll want to pay for it
English
0
0
0
15
Claude
Claude@claudeai·
New in Claude Cowork: teach Claude a skill. Record your screen while you do a task, talk through it as you go, and Claude turns it into a skill it can run again. Find it under Record a skill in the + menu of the Claude desktop app. Available on Pro, Max, and Team plans.
Claude tweet media
English
1.3K
3.3K
45.4K
10.9M
Cursor
Cursor@cursor_ai·
We had a team of agents rebuild SQLite from its 835-page manual. It created a replica in Rust which passed 100% of a held-out test suite. Interestingly, cost varied 15x depending on which model mix we used.
Cursor tweet media
English
607
901
12.8K
3.2M
Vox
Vox@Voxyz_ai·
one more thing: Orwell's rules for Claude Code/Codex missed the AI tic i recognize on sight: 𝗰𝗼𝗿𝗿𝗲𝗰𝘁𝗶𝘃𝗲 𝗷𝘂𝘅𝘁𝗮𝗽𝗼𝘀𝗶𝘁𝗶𝗼𝗻. the not X, it's Y shell: the answer isn't X, it's Y X isn't the problem, Y is same mold, almost any topic. my guess: it gives a plain statement the shape of insight without adding a fact. training supplies the pattern; post-training rewards the clean, decisive sound. i use it in normal conversation. the structure is fine. rhetoricians call the family 𝗮𝗻𝘁𝗶𝘁𝗵𝗲𝘀𝗶𝘀; the corrective form is 𝗰𝗼𝗿𝗿𝗲𝗰𝘁𝗶𝗼. JFK's ask not and MLK's color/content belong to it. rhetoricians used it for 2,000 years. LLMs wore it out in a few. wonder which good expression they'll ruin next lol. six more rules for CLAUDE.md / AGENTS.md. the 𝟮𝟬𝟮𝟲 𝗽𝗮𝘁𝗰𝗵: 7. don't build a straw man to knock down. use not X, it's Y once per piece, max 8. two examples are enough. don't stretch to three 9. don't announce what you're about to say. say it 10. don't end two paragraphs in a row with punchlines 11. vary the length and shape of neighboring sentences 12. break any of these rules sooner than write like a machine
Vox@Voxyz_ai

𝘀𝘁𝗼𝗽 telling Claude Code/Codex "no em dashes". 𝘀𝘁𝗼𝗽 telling Claude Code/Codex "stop saying delve". 𝘀𝘁𝗼𝗽 telling Claude Code/Codex "don't sound like AI". you never gave it a 𝘄𝗿𝗶𝘁𝗶𝗻𝗴 𝘀𝘆𝘀𝘁𝗲𝗺. every README, PR description and landing page ships in the same AI voice, and you ban words one at a time. orwell wrote the fix 𝟴𝟬 𝘆𝗲𝗮𝗿𝘀 𝗮𝗴𝗼. six rules, 1946. paste them into your global CLAUDE.md / AGENTS.md and every session picks them up: → never use a long word where a short one will do → if it is possible to cut a word out, cut it out → never use the passive where you can use the active 6 blocks you can copy-paste directly 👇

English
11
17
273
34.4K
typecraft
typecraft@typecraft_dev·
What's everyone's experience with herder? Looks pretty cool! herdr.dev
English
109
12
429
34K
Andre Carrera retweetledi
herdr
herdr@herdrdev·
herdr 0.7.4 is out! many fixes, many additions, and structural work for multi-client 👀 but this one ships one of the two most wanted features: customizable sidebar. both agent and workspace rows: plugins, custom scripts, agent hooks, or just settings can put anything on them. claude's live task title, codex context usage, a pac-man eating pi's context window. your agents, your rows.
English
47
91
929
90.5K
Elixir by Software Mansion
Elixir by Software Mansion@swmansionElixir·
Your @elixirphoenix LiveView app disconnects from the server. The UI freezes, nothing updates until the socket comes back. Local LiveView doesn't freeze, and when the connection returns, it doesn't throw away your state either. We built a small burrito ordering app with a "disconnect" button baked in – see how both versions handle the same disconnect live 🌯
English
8
9
94
4.8K
Andre Carrera
Andre Carrera@andr_ec_·
@AnthropicAI @ClaudeDevs trying to pay for a max plan then session shows as expired. Could you get your payments and verification fixed?
Andre Carrera tweet mediaAndre Carrera tweet media
English
0
0
0
32
Dane Knecht 🦭
Dane Knecht 🦭@dok2001·
Durable Objects are the best compute primitive for agents.
English
19
13
273
36.5K
Andre Carrera
Andre Carrera@andr_ec_·
@dan_note Do you have a solution for this in your pi harness? I’ve noticed no matter how clearly I specify it in AGENTS.md it still needs to be done in a second pass during review
English
1
0
0
37
Danila Poyarkov
Danila Poyarkov@dan_note·
All frontier models are still very bad at separating the conversational context from what should be retained in the code or shared with a third party — for example, the author of a PR you’re replying to — as well as what should remain private and what each party is already aware of.
zack@zack_overflow

Noticing Fable is very often writing code comments that reference details from chatting with it, but they make very little sense in the context of reading it in the codebase

English
2
0
8
839
super.engineering
super.engineering@superdoteng·
Horizontal tabs work well for a few conversations, but complex tasks can quickly outgrow them. Now you can see every tab at a glance, right in the sidebar.
English
10
9
218
23.2K
Andre Carrera
Andre Carrera@andr_ec_·
I feel like fable is the first model you can take advantage of the full 1M context window. even opus needs to be kept under 175k to not make silly mistakes.
Andre Carrera tweet media
English
1
0
1
33

Keşfet

@dexhorthy @mikehostetler @benhylak @claudeai @cursor_ai @Voxyz_ai @jamonholmgren @ThePrimeagen