Blog
Insights, analysis, and updates from the AI agent economy. Browse by tag · Browse the archive.
Coolify's v5 Rewrite Has No Public Timeline a Year In: What Betting on a Small-Team PaaS's Next-Gen Core Means for Production Users
Coolify's v5 rewrite promises cloud-like multi-server scaling, but its GitHub milestone sits at 0% completion a year after the first commit. Here's what the tracker actually shows and what it means if you're running production on Coolify today.
Coolify's MCP Server Ships Read-Only: Dokploy Beat It to AI-Driven Deploys, and Neither Can Answer 'Is My Fleet Healthy?'
Coolify's new MCP server can't deploy anything yet — it's 10 read-only tools. Dokploy shipped a write-capable one a month earlier. Here's what both still can't tell an AI agent about a multi-node fleet.
Coolify Shipped an MCP Server. It Still Can't Deploy Your App.
Coolify's official MCP server is strictly read-only — the write-capable 'AI deploys your app' story everyone cites comes from a third-party wrapper holding your API token. Here's the concrete tool-by-tool split, why it matters, and where bex's own MCP server stands on the same line.
Chaos Mesh vs LitmusChaos: Which One Actually Fits Breaking Your Own Fleet on Purpose
Chaos Mesh runs a privileged daemon on every node; LitmusChaos spins up a disposable pod per experiment. A concrete look at which one actually proves a self-hosted, multi-tenant Cluster API fleet survives a node loss, a network partition, and an etcd disk going slow — without taking down a tenant's app by accident.
Buildpacks vs the Self-Hosted PaaS Pack: Why Coolify, Dokploy, and Kubero All Default to Heroku's git-push Instead of Cloud Native Buildpacks
Heroku co-created the Cloud Native Buildpacks standard in 2018. In 2026, the leading self-hosted PaaS tools still don't default to it — and two of them support it anyway, just not by default. Here's the actual build layer under each tool's hood.
12-Factor Agents Is the New 12-Factor App: A Deploy Tool's Field Guide
Dex Horthy's 12-Factor Agents mirrors Heroku's 2011 methodology almost line for line. Here's what each of the twelve factors concretely costs a platform building MCP-exposed deploy and rollback tools for AI agents.
Vercel's Spending Cap Doesn't Cap Your Spending: The Bandwidth Math Behind $23K Bills
Vercel's Spend Management feature can pause a runaway bill, but it's opt-in, checks usage every few minutes instead of continuously, and shuts down all of production instead of capping the overage — here's the real per-GB, per-CPU-hour math and what a fixed-cost server costs instead.
Vercel's Active CPU Billing Saves 93% on a Real AI Agent — Here's the Exact Math
Vercel claims Active CPU pricing cuts costs up to 95% across 45 billion weekly requests. A worked example on a realistic AI-agent function gets 93% — and shows the exact invocation volume where the discounted bill catches back up to a flat-rate box you already own.
Talos Linux Meets Cluster API: The Immutable-OS Alternative to Ubuntu+kubeadm
Talos Linux strips the Kubernetes node OS down to a dozen binaries, no SSH, and a single mTLS API — and it plugs into Cluster API through its own bootstrap and control-plane providers, right alongside CAPH on Hetzner.
Subscribe
New posts land in your reader as soon as they publish. Pick a format — all three carry the same posts.
Current feeds keep roughly two days of posts so daily polling does not miss a burst. Older entries stay reachable from the feed's next-page link in readers that follow it, or from the blog archive.
Following one topic instead? Browse tags