Code is clay now

Artisans, factories and the new material software is made of

Using LLM for confidence scoring is a bad idea

LLM-based agents can do a lot of tasks. Or, at least, they can appear to do them. For example, they write code, summarize papers, plan trips, and fill in forms. If you ask one for a probability or a confidence score, it will give you one of those too. I have seen many agent designers use LLMs to generate confidence scores in this way. The problem is that there is no good reason to trust that number. This post is about why we cannot rely on an LLM to generate a confidence score.

No more icons in every menu by default

Have you heard the good news? I bemoaned the approach of putting an icon in every menu item by default which shipped with macOS Tahoe — but now it’s gone in…

Security Disclosure

Railway detected anomalous activity involving systems that store customer environment variables and recommends your immediate attention

LLM agents play Pokemon showdown

Two LLM agents share a side in 4-player Pokémon Showdown battles and can talk before every turn. An open environment for measuring whether natural-language coordination wins more, and what it costs.

Beyond the &

Chain of Thought for Humans

There isn't an easy exploit for this PHP vulnerability

Mass hysteria, or something.

Today's PHP security release, and how we're strengthening security

Describing the story behind how critical security vulnerabilities were found and fixed.

Clockwork Mechanics, Maintenance, and Metaphors in the Late Middle Ages

I can't believe PowerShell is good

PowerShell, I owe you an apology. I wasn't really familiar with your game.

PingCap TiDB Leaks

GitLab critical patch release

Learn more about GitLab Critical Patch Release: 19.5.1, 19.4.3, 19.3.6, 19.2.9, 19.1.9, 19.0.10, 18.11.13 for GitLab Community Edition (CE) and Enterprise Edition (EE).

The recently proposed o(n log n) integer multiplication algorithm [pdf]

Rust Style Guide

Style guides for Google-originated open-source projects

Compromised LiteLLM 1.107.0 on PyPI Installs a Credential Stealer and Backdoor

LiteLLM 1.107.0 on PyPI ships a .pth credential stealer and a sitecustomize.py backdoor that run on every Python start. How the implant got in, what it steals, and indicators of compromise.

Show HN: Omnara – Open-source alternative to Claude Managed Agents

The open-source managed agent platform. A self-hostable alternative to Claude Managed Agents and OpenAI's Agents API. - omnara-ai/omnara

Tiny possum and glider thought extinct for 6000 years found in remote West Papua

The discovery of two lost species is 'exceptional' and rare, scientists say

Meta ordered to pay $25B penalty in New Mexico Cambridge Analytica case

Meta to pay $25 billion in penalties related to Cambridge Analytica scandal in New Mexico.,Meta to pay $25 billion in penalties related to Cambridge Analytica scandal in New Mexico.

mitmproxy adds --emudhra option

An interactive TLS-capable intercepting HTTP proxy for penetration testers and software developers. - Add `--emudhra` option (#8500) · mitmproxy/mitmproxy@1692344