Daily Reading List – August 13, 2025 (#608)

After yesterday’s MCP debacle, I built a new server using FastMCP and Python. All was right with the world. It was fun to create an API, expose it through tools in an MCP server, and then have my agent (the Gemini CLI) use those tools to invoke the API. I found it useful to build it all out!

[blog] Scaling down to speed up: How we improved efficiency of live process metrics by 100x. Datadog shares what they did, and it may inspire your own efforts to improve timeliness of data.

[blog] Gemini CLI Tutorial Series — Part 9: Understanding Context, Memory and Conversational Branching. Terrific post here on how to reset context, save and branch sessions, and other tips for our agentic CLI.

[blog] Gemini CLI + VS Code: Native diffing and context-aware workflows. This is big. I installed it earlier today, and it’s so handy to have a souped-up terminal when I’m working in the IDE.

[blog] AI Prototyping: How 11 Real-World Teams Are Transforming Their Work with Lovable. What are product people doing with AI prototyping tools like Lovable? I like seeing these real stories.

[blog] Secure your LLM apps with Google Cloud Model Armor. You’ll find a few ways to protect your LLM apps. This is an underrated service that can detect prompt injection attempts, block malicious URLs, and prevent sensitive data leakage.

[article] Unpacking METR’s findings: Does AI slow developers down? Some fresh perspectives from someone who participated in this buzzy study.

[blog] How to Stop Unauthorized API Access. Trying to keep people out of your APIs? This is a pretty good list of ways to do so.

[blog] Still Packaging AI Models in Containers? Do This Instead on Cloud Run. Powerful use case from Karl. You can stream the model files and keep it separate from your app code.

[blog] Go 1.25 is released. Nothing dramatic in this release, but some useful upgrades to tools, compilation, standard libraries, and garbage collection. Release notes here.

[blog] Gemini CLI or Jules: Which tool is for you? Great explainer. Yes, many tools have overlapping scenarios, even when coming from the same creator. But there are distinctions that matter.

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

Comments

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.