camar 

Love life minus flu & liars, expect parody, satire & sarcasm

[Daily Dose of Data Science @DailyDoseOfDS_]

Claude Code fully dissected!

Researchers from UCL reverse-engineered the leaked Claude source. What they found changes how you should think about agent design.

Only 1.6% of the codebase is AI decision logic.

The other 98.4% is operational infrastructure. Permission gates, tool routing, context compaction, recovery logic, session persistence. The model reasons. The harness does everything else.

This is the opposite of what most agent frameworks do today.

LangGraph routes model outputs through explicit state machines. Devin bolts heavy planners onto operational scaffolding. Claude Code gives the model maximum decision latitude inside a rich deterministic harness, and invests all its engineering effort in that harness.

The core loop is a simple while-true. Call model, run tools, repeat.

But the systems around that loop are where the real design lives:

A permission system with 7 modes and an ML classifier. Users approve 93% of prompts anyway, so the architecture compensates with automated layers instead of adding more warnings.

A 5-layer context compaction pipeline. Each layer runs only when cheaper ones fail. Budget reduction, snip, microcompact, context collapse, auto-compact.

Four extension mechanisms ordered by context cost. Hooks (zero), skills (low), plugins (medium), MCP (high). Each answers a different integration problem.

Subagents return only summary text to the parent. Their full transcripts live in sidechain files. Agent teams still cost roughly 7x the tokens of a standard session.

Resume does not restore session-scoped permissions. Trust is re-established every session. That friction is the point.

The bet behind all of this is simple. As frontier models converge on raw coding ability, the quality of the harness becomes the differentiator, not the model.

Paper: Dive into Claude Code (arXiv:2604.14228)

We've shared an article on Agent Harness and what every big company is building.

Read it below.

[camar  @camarmir]

@SouthAsiaIndex US Administration has never been known for admissions, let alone apologies for the screw ups. Godi Media & Modi will have to suck it up as a small token for its diplomatic positioning as an ally.

[Marie Haynes @Marie_Haynes]

This is really big news. Google introduced the Open Knowledge Format (OKF) - a standardized way to store information in a directory of markdown files. Makes it really easy to make a digital brain that agents can use.

These files can serve as a living wiki. You can give agents the ability to query them or edit them. They can interlink.

Seems to me this could replace Notion or Obsidian. I can think of so many uses for this.

Google's blog post: https://t.co/DqSjg4UpvH

An easier to understand explanation is the SPEC.md file:
https://t.co/A3qSz3Tfas

I gave those two links to Antigravity and asked how we could use it for any of the projects we're working on. It came up with so many ideas. I would imagine Claude Fable 5 would whip up some pretty amazing things based on this system.

Currently creating an OKF library of our pepper garden. It's going to be a fun weekend.

[Charly Wargnier @DataChaz]

🚨 @Karpathy predicted the power of the "LLM Wiki." Google just formalized it.

Meet Open Knowledge Format (OKF): a vendor-neutral standard for giving foundation models the curated context they need.

I can genuinely see this replacing Notion, Obsidian, or traditional wikis for developer teams, and the reason comes down to bookkeeping.

Traditional wikis fail because humans inevitably abandon the tedious work of updating them.

As Andrej Karpathy pointed out recently, LLMs don't get bored.

They don't forget to update a cross-reference, and they can touch 15 files in a single pass.

OKF standardizes the interoperability layer so agents can actually do that heavy lifting autonomously.

Because the format is minimally opinionated, it doesn't dictate what you write, it just dictates how it's structured. You get:
→ Human-readable documents that live right alongside your code in version control
→ Cross-links that map out complex entity relationships without needing a graph database
→ A system that survives moving between different tools and organizations

There is no complex compression scheme.

No central registry.

If you can cat a file, you can read it.

If you can git clone a repo, you can deploy it.

This is how we stop rebuilding context pipelines from scratch every time a new model drops.

Announcement + spec file in 🧵↓

[Trita Parsi @tparsi]

I have fought the neocons and warmongers in Washington for more than 25 years. Throughout, they have tried to silence, discredit, slander, and cancel me. Only recently, however, have they tried to deport me.

At least, that appears to have been the aim of a hit piece in Bari Weiss’s The Free Press, which claimed that Marco Rubio’s State Department was “investigating” me for allegedly seeking to “undermine the U.S.”—presumably because of my opposition to war with Iran.

Yet just hours later, the State Department issued a statement to reporters clarifying that “the State Department has no plans to revoke the green card of Mr. Parsi at this time.” Nor did it provide any confirmation for the central premise of the Free Press story—that an investigation of me existed in the first place.

So here’s what I think happened.

Read the full piece on my Substack: https://t.co/bjh5aEoLnL

[Sonny @rawespresso]

Owning a £400,000 house with a £350,000 mortgage in the UK in 2026 means signing a 25-year contract to give the bank around £2,050 a month, in exchange for the right to maintain a building you'll spend another £4,000 a year keeping dry, warm and structurally sound.

At a 5% fixed rate, the total interest over the life of the mortgage is around £264,000. The total mortgage cost ends up at roughly £614,000.

The price tag on the front door says £400K. The actual cost of living in it for 25 years is closer to £750K, before council tax, repairs, insurance and the new boiler.

This is what the country has been calling 'getting on the ladder.'

[International Cyber Digest @IntCyberDigest]

‼️The ban on foreign use of Anthropic's frontier models, might also cripple the development of USA's next-gen models.

Anthropic's stated the order suspends access by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.

[camar  @camarmir]

@RnaudBertrand A terrible PR / statement with ZERO empathy to the human losses (168 school children in Menab) where Claude played “A role”. He’s using it as pretext for potentially 100x worse outcomes in place that’d “matter”. Imagine these richest bigots owning US Military & Kill Machines.

[Arnaud Bertrand @RnaudBertrand]

The worst part is him saying the strike on the elementary school in Minab (that killed 168 schoolchildren and teachers) "is a use case that doesn't even violate our red lines."

[International Cyber Digest @IntCyberDigest]

‼️🚨 BREAKING: Amazon researchers snitched to the US government about jailbreaking Fable 5 and Mythos 5, forcing Anthropic to immediately shut down worldwide access.

A security export control directive from Commerce Secretary Howard Lutnick enforced the action.

Anthropic is fighting the directive and calls it a misunderstanding.

This isn't the first clash. The Trump administration had already tried to get Anthropic to pause the release of its latest models before this directive landed.

[Chubby♨️ @kimmonismus]

Wait - so Amazon, one of Anthropic’s biggest investors, allegedly jailbroke Claude and then snitched to the U.S. government?

This cant be real. What.

[محمدباقر قالیباف | MB Ghalibaf @mb_ghalibaf]

Commitments made must be commitments kept. No ifs, no buts, no excuses. For the close deal ahead, there is no other way.
You reap what you sow.

[High Yield Harry @HighyieldHarry]

Inspiring.

This young man came to America with no hair and zero bitches.

Now he has a full head of hair and 55 children.

[IPO Newsroom @IPONewsroom_]

The SpaceX $SPCX IPO has received $100+ Billion of orders from retail investors

SpaceX is looking to raise $75 Billion in the biggest IPO in history

The biggest IPO in history could be fulfilled by JUST retail investors

[Peter Schiff @PeterSchiff]

With the SpaceX IPO @elonmusk is the first person to surpass J.D. Rockefeller as the richest private citizen ever. However, adjusted for over 100 years of inflation, Rockefeller earned about $2 billion per year in real income. Musk earns nothing. What a difference a bubble makes.

[Big Brain AI @realBigBrainAI]

Google DeepMind CEO Demis Hassabis on the central question of his life: what can humans do that AI never will?

Asked about the limits of artificial intelligence, Hassabis doesn't hesitate to say this is the question that drives him: "Yes, it is."

His starting point is one of his heroes, Alan Turing.

He explains that Turing described theoretical constructs—Turing machines—that underpin all modern computers, machines "that are able to compute anything that's computable… anything that can be described as an algorithm."

The provocative leap comes when Hassabis turns that lens on biology. The AI systems his teams build are Turing machines, and he suspects the brain may be one too:

"A lot of neuroscientists including me think that maybe the brain… a good model for the brain is an approximate Turing machine."

He's careful to note this isn't settled. Some of his peers see it differently, including physicist Roger Penrose:

"Friends of mine like Roger Penrose… believe there might be some quantum effect in the brain."

But Hassabis points to where the evidence currently stands:

"So far neuroscience hasn't found any quantum effects in the brain… people have looked quite carefully and we haven't found any."

His conclusion from that absence is striking. If the brain runs on ordinary computation, then there may be no hard ceiling on what AI can eventually match:

"It looks like most of what's going on in the brain is kind of classical computation… so therefore it's not clear what the limit would be in terms of eventually what an AI system could do and could mimic."

For @demishassabis, building intelligence is a mirror held up to ourselves, not only a feat of engineering.

He describes the project as a kind of experiment that reveals what we are:

"I think we'll have almost like a control study comparison to the human mind. And then I think we'll see in this journey what are the differences and what's unique about the mind."

He stays genuinely open about what that comparison might expose. Some things, he suspects, may never transfer:

"There could be unique things and certainly unique connections between humans that will never be replicated by these AI systems."

But the capabilities we tend to treat as distinctly human, he believes, are within reach:

"A lot of things that we currently are not in reach, like long-term planning and reasoning and maybe some forms of creativity... I think eventually AI systems will be able to do."

[Dr.Sam Youssef Ph.D.,Ph.D.,DPT. @drhossamsamy65]

⛔️ Virginia Giuffre revealed Epstein trafficked her to Andrew. She’s hit by a bus then “commits suicide.”‼️

⛔️The interviewer’s mother gets kidnapped and is still being held for ransom.‼️

⛔️Can you connect the dots ‼️

[camar  @camarmir]

@TheSecretAcct Given the appetite in retail for it, it will cross $2 Trillion before the month end and $3 Trillion before EOY