The maintainer of Backpack for Laravel, a decade-old admin panel package, reports that tasks once taking a day or two now take about an hour thanks to AI coding agents. They note that two widely recommended tools for making packages 'AI-ready' — an llms.txt file and an MCP server — provided little measurable benefit, as agents rarely used either effectively.
backpackforlaravel.com
· 2026-11-20
A software engineering blog post argues that developers who stop writing code themselves—relying instead on instructing AI or others and reviewing the output—will see their underlying skills decline, just as musicians or authors would lose ability if they stopped practicing their craft. The author challenges the industry assumption that one can maintain or improve coding expertise purely through specifying requirements and giving feedback on others' output.
cekrem.github.io
· 2026-11-20
A developer used Anthropic's Claude Code to rebuild his abandoned game KinForge, converting it from AWS-hosted servers to peer-to-peer play with persistent save states in about four hours. Epic Games COO Daniel Vogel acknowledged that AI models capable of generating custom game engines and code, such as Anthropic's Fable 5.2, are pushing Epic to increase the pace of Unreal Engine development.
spectrum.ieee.org
· 2026-11-20
The PHP project released a security update on Nov. 19 fixing dozens of bugs, including seven rated critical (CVSS 9.2) that could enable remote code execution or request smuggling under certain configurations. Patches cover versions 8.2.36, 8.3.37, 8.4.28, 8.5.13, and the newly launched 8.6.0, with maintainers urging prompt upgrades.
darkreading.com
· 2026-11-19
An AI engineer built Rune Score, a sports analytics chatbot, and ChatMakr, a platform for generating multi-agent chatbots, without paying for any tools. The stack spans branding, code hosting, and development, including Figma for logo design and GitHub for repository hosting and CI minutes, each used within free-tier limits.
entrepreneur.com
· 2026-11-19
Onlook, part of YC's W25 batch, has launched an open-source desktop tool that lets designers visually edit a live, running version of their actual codebase rather than a mockup. The tool doesn't include its own AI model; instead, users connect existing Claude Code or ChatGPT/Codex accounts, which write the code while designers shape the interface on a canvas. It works directly with local Git repositories and dev servers, and supports MCP connectors for additional context.
github.com
· 2026-11-19
A product engineer reports that Anthropic's Claude Code refused to help implement a one-line security fix meant to patch a vulnerability in their own product, citing AI safety restrictions. The author says Anthropic's process for requesting an exemption requires submitting government ID, waiving data retention protections, and waiting multiple business days, with no guarantee of approval.
ounapuu.ee
· 2026-11-19
The PHP development team released security updates on November 19 covering versions 8.5.13, 8.4.28, 8.3.37, and 8.2.36, fixing 18 CVEs across the core engine, session handling, PHP-FPM, and CGI components. Security firm Aikido flagged the most severe as CVE-2026-129289, a critical PHP-FPM request smuggling flaw with a CVSS score of 9.2 that could let unauthenticated attackers remotely execute code on servers reusing backend connections to PHP-FPM.
bleepingcomputer.com
· 2026-11-19
Security researchers at Calif Research disclosed a critical vulnerability in OpenJDK, the base library underlying most Java implementations, that enables remote code execution often without authentication. The flaw lies in a Base64 decoding optimization and has been rated between 8 and 9.8 on CVSS scales by various vendors, including Oracle and Red Hat. Oracle has released patched versions, but the researchers demonstrated working exploits against widely used software including Keycloak, Jira, Confluence, Kafka and Elasticsearch.
tomshardware.com
· 2026-11-19
Researchers from Xint, Wiz, Ikotas Labs, Metnir and Rapid7 independently found two critical HTTP/2 vulnerabilities in NGINX that let unauthenticated attackers run code remotely, with one chainable to a Linux kernel SCTP bug for full root access. Xint also found three more use-after-free flaws affecting HTTP/3, DNS resolution and mail proxy authentication. F5 has patched all five issues in NGINX Open Source 1.30.7 and 1.31.9.
bleepingcomputer.com
· 2026-11-19
The Drupal Security Team released patches for a critical remote code execution vulnerability (CVE-2026-132165) that lets anonymous users abuse the CMS's install.php file to reinstall Drupal and potentially take over a site. The flaw, found by Anthropic's OSS Scanner, was fixed in Drupal 10.6.22, 11.3.22, and 11.4.13, alongside a separate moderately critical deserialization vulnerability (CVE-2026-132166) requiring admin access.
bleepingcomputer.com
· 2026-11-19
Security firm Calif discovered a critical Oracle Java vulnerability, nicknamed Java4Shell, that corrupts memory during Base64 decoding and could let attackers run malicious code remotely without a username or password. Oracle has issued a patch after being notified in October, and CISA is now urging organizations to apply it. Oracle rated the flaw 8.1 out of 10 in severity.
cnet.com
· 2026-11-19