Code Centre

A vibe coding enthusiast. Quality over quantity. Pro tip: click on the Highlights tab to see the most useful posts

[Code Centre @codecentrevibe]

#vibecoding #claudecode
https://t.co/1dO0dJJX8u

[Elon Musk @elonmusk]

The scale of what is to come has no precedent

[ClaudeDevs @ClaudeDevs]

For the developers building with Claude, a direct line from the team.

Follow for changelogs, API releases, community updates, and deep dives.

[Claude @claudeai]

Introducing Claude Opus 4.7, our most capable Opus model yet.

It handles long-running tasks with more rigor, follows instructions more precisely, and verifies its own outputs before reporting back.

You can hand off your hardest work with less supervision.

[Thariq @trq212]

you'll need to explicitly prompt Claude Code to use it, but the Monitor Tool is super powerful

e.g. "start my dev server and use the MonitorTool to observe for errors"

[Andrej Karpathy @karpathy]

LLM Knowledge Bases

Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So:

Data ingest:
I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them.

IDE:
I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides).

Q&A:
Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale.

Output:
Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base.

Linting:
I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into.

Extra tools:
I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries.

Further explorations:
As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows.

TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

[Paweł Huryn @PawelHuryn]

Plan mode isn't a planning tool. It's a feedback guarantee.

You share your intent. The agent shows what it understood. You confirm before it executes. No ambiguity about whether it's listening or already coding.

The military has a name for this: Commander's Intent. Give the end state and reasoning upfront, then let the team run autonomously. The alignment step is what makes the autonomy work.

OpenClaw doesn't have plan mode yet. Curious how long that lasts.

[Boris Cherny @bcherny]

Today we're excited to announce NO_FLICKER mode for Claude Code in the terminal

It uses an experimental new renderer that we're excited about. The renderer is early and has tradeoffs, but already we've found that most internal users prefer it over the old renderer. It also supports mouse events (yes, in a terminal).

Try it: CLAUDE_CODE_NO_FLICKER=1 claude

[Thariq @trq212]

not an April Fools joke, we rewrote the Claude Code renderer to use a virtual viewport

you can use your mouse, the prompt input stays at the bottom, and a lot more small UX wins people have been asking for

it's experimental so give us your feedback

[Claude @claudeai]

New in Claude Code: auto mode.

Instead of approving every file write and bash command, or skipping permissions entirely, auto mode lets Claude make permission decisions on your behalf.

Safeguards check each action before it runs.

[Claude @claudeai]

You can now enable Claude to use your computer to complete tasks.

It opens your apps, navigates your browser, fills in spreadsheets—anything you'd do sitting at your desk.

Research preview in Claude Cowork and Claude Code, macOS only.

[Thariq @trq212]

We just released Claude Code channels, which allows you to control your Claude Code session through select MCPs, starting with Telegram and Discord.

Use this to message Claude Code directly from your phone.

[Thariq @trq212]

https://t.co/45C3gKydTK

[David Geoghegan @Dave_Geoghegan_]

@trq212 the workflow changes completely when you stop re-explaining context every few messages. feels like working with someone who was actually in the room the whole time

[Andrej Karpathy @karpathy]

Three days ago I left autoresearch tuning nanochat for ~2 days on depth=12 model. It found ~20 changes that improved the validation loss. I tested these changes yesterday and all of them were additive and transferred to larger (depth=24) models. Stacking up all of these changes, today I measured that the leaderboard's "Time to GPT-2" drops from 2.02 hours to 1.80 hours (~11% improvement), this will be the new leaderboard entry. So yes, these are real improvements and they make an actual difference. I am mildly surprised that my very first naive attempt already worked this well on top of what I thought was already a fairly manually well-tuned project.

This is a first for me because I am very used to doing the iterative optimization of neural network training manually. You come up with ideas, you implement them, you check if they work (better validation loss), you come up with new ideas based on that, you read some papers for inspiration, etc etc. This is the bread and butter of what I do daily for 2 decades. Seeing the agent do this entire workflow end-to-end and all by itself as it worked through approx. 700 changes autonomously is wild. It really looked at the sequence of results of experiments and used that to plan the next ones. It's not novel, ground-breaking "research" (yet), but all the adjustments are "real", I didn't find them manually previously, and they stack up and actually improved nanochat. Among the bigger things e.g.:

- It noticed an oversight that my parameterless QKnorm didn't have a scaler multiplier attached, so my attention was too diffuse. The agent found multipliers to sharpen it, pointing to future work.
- It found that the Value Embeddings really like regularization and I wasn't applying any (oops).
- It found that my banded attention was too conservative (i forgot to tune it).
- It found that AdamW betas were all messed up.
- It tuned the weight decay schedule.
- It tuned the network initialization.

This is on top of all the tuning I've already done over a good amount of time. The exact commit is here, from this "round 1" of autoresearch. I am going to kick off "round 2", and in parallel I am looking at how multiple agents can collaborate to unlock parallelism.
https://t.co/WAz8aIztKT

All LLM frontier labs will do this. It's the final boss battle. It's a lot more complex at scale of course - you don't just have a single train. py file to tune. But doing it is "just engineering" and it's going to work. You spin up a swarm of agents, you have them collaborate to tune smaller models, you promote the most promising ideas to increasingly larger scales, and humans (optionally) contribute on the edges.

And more generally, *any* metric you care about that is reasonably efficient to evaluate (or that has more efficient proxy metrics such as training a smaller network) can be autoresearched by an agent swarm. It's worth thinking about whether your problem falls into this bucket too.

[Dickson Tsai @dickson_tsai]

In Claude Code, we’ve recently launched HTTP hooks, easier to use and more secure than existing command hooks! You can build a web app (even on localhost) to view CC’s progress, manage its permissions, and more. Then, now that you have a server with your hooks processing logic, you can easily deploy new changes or manage state across your CCs with a DB.

How do HTTP hooks work? CC posts the hook event to a URL of your choice and awaits a response. They work wherever hooks are supported, including plugins, custom agents, and enterprise managed settings.

Docs: https://t.co/ihQWcpOlGA

[Thariq @trq212]

a few Friday afternoon ships to end the week:

the AskUserQuestion tool can now show markdown snippets to display diagrams, code examples, etc.

[Noah Zweben @noahzweben]

Rolling out Claude Code Remote Control to Pro users - because they deserve to use the bathroom too .
(Team and Enterprise coming soon).

🧻 Rolling out to 10% and ramping

1. Update to claude v2.1.58+
2. Try log-out and log-in to get fresh flag values.
3. /remote-control

[Thariq @trq212]

We've rolled out a new auto-memory feature.

Claude now remembers what it learns across sessions — your project context, debugging patterns, preferred approaches — and recalls it later without you having to write anything down.

[Thariq @trq212]

https://t.co/nKTDfC7zMm

[Y Combinator @ycombinator]

A very special guest on this episode of the Lightcone! @bcherny, the creator of Claude Code, sits down to share the incredible journey of developing one of the most transformative coding tools of the AI era.

00:00 Intro
01:45 The most surprising moment in the rise of Claude Code
02:38 How Boris came up with the idea for Claude Code
05:38 The elegant simplicity of terminals
07:09 The first use cases
09:00 What’s in Boris’ https://t.co/OAtnXdxccP?
11:29 How do you decide the terminal’s verbosity?
15:44 Beginner’s mindset is key as the models improve
18:56 Hyper specialists vs hyper generalists
21:51 The vision for Claude teams
23:48 Subagents
25:12 A world without plan mode?
28:38 Tips for founders to build for the future
30:07 How much life does the terminal still have?
30:57 Advice for dev tool founders
32:11 Claude Code and TypeScript parallels
35:34 Designing for the terminal was hard
37:36 Other advice for builders
40:31 Productivity per engineer
41:36 Why Boris chose to join Anthropic
44:46 How coding will change
46:22 Outro