Daily Reading List – August 3, 2026 (#837)

I’m back today with a big reading list. The short weekend in Seattle was awesome, and I got to see some people I’ve missed since moving away four years ago. I also missed Taco Time, and had to stop there. No apologies.

[blog] Batten Down Your Packages: Mitigation Guidance for Supply Chain Compromise. Learn more about supply chain compromises, and how to protect yourself against this ever-present risk.

[blog] Modern Software Registries are a Trust Service. Related to above. Identity and provenance play a huge part in establishing (and maintaining) trust.

[blog] What happens to a lawyer’s business model when AI makes him 5x faster. There’s a giant set of new “builders” out there. How much will they build? Will they build regularly? Do they have loyalties to a specific stack? I’m not sure anyone knows, but don’t ignore this constituency.

[article] The Economic Benefit of Refactoring. If you have less code, you spend less money on tokens when your AI tool reads it. This is an example of refactoring some bloated AI-generated code and seeing the benefits.

[blog] the one line 95% of agent skills are missing. We’re skipping the instruction that tells the AI tool when an agent should load them.

[blog] From ‘Write a Python Script’ to Multi-Agent Mastery: My Journey to Becoming an AI Builder. We can build in areas outside our core knowledge areas. Amanda might not have been an audio/visual app developer, but that didn’t stop her.

[article] Research: How AI Agents Broaden the Scope of Knowledge Work. It’s not just about doing the current work better. It’s about different work.

[blog] Behind the scenes: How we build, test, and scale Google Agent Skills. How do you create skills at scale? I’m very proud of my team’s work here, and the engineering discipline they put into the effort.

[blog] How Agentic Coding Is Reshaping the Software Development Lifecycle. Very interesting lens on the new SDLC. Are you building for the always-on PM? Software factories that store, build, and secure code differently?

[blog] Do more with less: How GKE can reduce your cost per agent by 75%. From 61 OpenClaw agents on a vanilla Kubernetes node to 274 with a cost optimized, high-density configuration. Pretty good!

[blog] Giving and taking credit in big tech companies. Do you feel a little weird loudly taking credit for something? Probably, unless you’re a narcissistic psycho. But remember that you need to both take and give credit generously in a corporate environment.

[blog] The borderless Lakehouse: Bring AWS, Databricks and Snowflake data to your AI agents. Looks legit. Understand and query data regardless of its physical location.

[blog] Stronger with every update: How we’re making Chrome and the web safer in the AI Era. Excellent post that offers lessons for anyone trying to improve product quality.

[blog] Who’s Writing Open Source Code? Very interesting analysis. Are AI robots writing all the committed open source code now? Not remotely the case (yet).

[blog] Adapting open source practices to an AI-first world: A retrospective on 2025. Related to the previous one, thousands of Googlers are actively contributing to open source projects.

[blog] TypeScript Just Got 10x Faster by Not Being TypeScript. Go was a smart choice by this team. Performance is great, and Go suited the compiler’s needs.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below: