Almost all of the work I delegate to agents is work I could have done myself. The agent gives me two things: more thoroughness (which is now always better) and faster completion. I could sit down on a quiet day (which never happens) and type up a technical design doc for the team. I could dig through the logs and metrics correlating the latency bump from overnight. I could grep the codebase for the error message and trace through the code manually.

I believe one of the reasons I’ve been pretty successful in adopting and utilizing Agents iny engineering work is relying on the decades of hard fought knowledge, experience, and taste. How do folks without that get it in this age of AI? It’s unacceptable for me to say “AI is only for seniors and above” or other nonsensical takes. Here’s what I’m currently thinking. And as with most everything “rethought” for the AI era, it’s the same as before.

System Design used to be reserved for the senior folks. They’ll hand down a document showcasing the new architecture or new service. Agents are making it so everyone can do some system design because even the smallest features can benefit from more intentional design.

AI system-design

Agentic coding harnesses are violating the unix philosophy and trying to become the everything app for coding (and work and personal assistant and so on).

AI Thoughts

A former coworker of mine over 8 years ago put this image into my head that has never left. He was working as my manager at the time and he had an upcoming large, foundational shifting feature that he wanted to let me know was coming. He said “Don’t do anything right now, but just put this in your noodler and see what comes out.”

Thoughts Slow then fast

golibnotify is a Go library for sending desktop notifications on Linux. It is a wrapper around the libnotify library.

There are some memory leak fixes in the latest release, v0.2.0.

Golang Open Source My Stuff

I often get large JSON files which contain no newlines or indentation. I’d like to look at these reasonably in neovim, but it’s been cumbersome until today.

TL;DR: You can replace the contents of the buffer with output formatted by jq with one command: :%!jq .

TIL

How can a tool that is indescernably wrong be anything more than a toy?

large-language-models

Here’s a super quick one-liner to copy the current branch’s PR URL to your clipboard (for posting into Slack, tickets, etc.) $ gh pr view --json url --jq .url | <pbcopy | xclip -sel c> Remember to replace the copy command with your platform’s specific command. This would also make a good shell alias. Also, also, I might try setting up Clipboard for multi-platform goodness. Happy PR-ing! – Chris

git Command-Line

With the wide adoption of robust front-end frameworks and CSS-in-JS, traditional methods of targeting xpath or CSS selectors for scraping data from web pages are becoming ineffective. However, this new paradigm of front-end development leaves a key opportunity to improve the web scraping experience.

Thoughts