<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Posts on idodo.dev</title><link>https://idodo.dev/posts/</link><description>Recent content in Posts on idodo.dev</description><generator>Hugo -- gohugo.io</generator><language>en-us</language><lastBuildDate>Sun, 13 Sep 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://idodo.dev/posts/index.xml" rel="self" type="application/rss+xml"/><item><title>entra-tui: Microsoft Entra ID, Now Without the Mouse</title><link>https://idodo.dev/posts/entra-tui/</link><pubDate>Sun, 13 Sep 2026 00:00:00 +0000</pubDate><guid>https://idodo.dev/posts/entra-tui/</guid><description>I keep building terminal UIs, because the mouse is a peripheral I resent. entra-tui is the newest one: a k9s-flavoured browser for Microsoft Entra ID that borrows your az login token, never holds one of its own, and gets you to an object in three keystrokes instead of nine clicks. Written entirely in Go — by Claude, with me driving.</description></item><item><title>TFTUI 0.14: I Handed My Own Project to an AI for a Day</title><link>https://idodo.dev/posts/tftui-0-14/</link><pubDate>Mon, 07 Sep 2026 00:00:00 +0000</pubDate><guid>https://idodo.dev/posts/tftui-0-14/</guid><description>TFTUI is a terminal UI for Terraform state that I wrote three years ago, mostly by feel. Version 0.14 is a complete rewrite — new internals, 406 tests, CI on three platforms — that Claude Code did in a single day while I drove and reviewed. Here is what changed, what broke, and what the experience actually taught me.</description></item><item><title>Apigee Spike Arrest: How the Bucket Actually Works</title><link>https://idodo.dev/posts/apigee-spikearrest/</link><pubDate>Fri, 14 Aug 2026 00:00:00 +0000</pubDate><guid>https://idodo.dev/posts/apigee-spikearrest/</guid><description>What a 429 means, why equivalent throughput rates can behave differently, and how clients should respond when Apigee SpikeArrest rejects a request.</description></item><item><title>GitHub Repo Watcher: a quieter way to watch your repos</title><link>https://idodo.dev/posts/github-repo-watcher/</link><pubDate>Thu, 30 Jul 2026 00:00:00 +0000</pubDate><guid>https://idodo.dev/posts/github-repo-watcher/</guid><description>&lt;p>GitHub&amp;rsquo;s notification inbox has a design philosophy: tell you everything, and let
you sort it out. Someone labels an issue, a bot bumps a dependency, a colleague
replies to a thread you commented on eleven months ago — it all lands in the same
pile. So most of us do one of two things. We check the pile constantly, or we stop
checking it at all.&lt;/p>
&lt;p>&lt;strong>GitHub Repo Watcher&lt;/strong> takes a narrower swing. It watches a handful of repos you
actually care about and answers one question from your browser toolbar: &lt;em>is
anything waiting on me right now?&lt;/em>&lt;/p>
&lt;p>Most of the time the answer is a small green dot. When it isn&amp;rsquo;t, you get a number.&lt;/p></description></item><item><title>Broadcasting Video on AWS: Architecture, Options, and Cost Math</title><link>https://idodo.dev/posts/aws-streaming-architecture/</link><pubDate>Mon, 13 Apr 2026 07:51:27 +0300</pubDate><guid>https://idodo.dev/posts/aws-streaming-architecture/</guid><description>A technical deep-dive into how bits move from encoder to viewer — and what it costs.
Introduction Streaming video at scale is, at its core, a bandwidth and caching problem. Every architectural decision — which AWS services to use, how long your segments are, how much delay you&amp;rsquo;re willing to tolerate — has a direct, calculable effect on your monthly bill. For a hands-on way to estimate these numbers, check out our interactive AWS Streaming Cost Dashboard.</description></item><item><title>Vibe-Coding the Last Mile</title><link>https://idodo.dev/posts/vibe-coding-the-last-mile/</link><pubDate>Thu, 26 Feb 2026 20:39:23 +0200</pubDate><guid>https://idodo.dev/posts/vibe-coding-the-last-mile/</guid><description>Sometimes you don&amp;rsquo;t need to conquer the world or reinvent the wheel — you just need to add the final touch on an already working tool with a blind spot.
The Premise There&amp;rsquo;s a certain kind of problem that doesn&amp;rsquo;t get talked about enough in platform engineering: the almost great tool.
You&amp;rsquo;ve procured it. It&amp;rsquo;s enterprise-grade. It solves the core problem. It has a support contract and a roadmap and a Gartner Magic Quadrant placement.</description></item><item><title>How I Slashed an Entire Org’s GitHub Actions Spend by 50% with One Repo</title><link>https://idodo.dev/posts/github-workflows-spend-reduction/</link><pubDate>Sat, 24 Jan 2026 16:19:19 +0200</pubDate><guid>https://idodo.dev/posts/github-workflows-spend-reduction/</guid><description>In an organization with hundreds of repositories, you wouldn’t expect a single workflow in a single repo to be the primary driver of cost. Yet, in our case, the Pareto principle hit hard: one high-traffic repository was responsible for a staggering 50% of our entire GitHub Actions spend.
This is a story about leverage. Most engineering teams try to shave 5% off everything; we saved 50% of the entire org&amp;rsquo;s bill by identifying and fixing the right 1%.</description></item></channel></rss>