Claude Opus 5.5 for WordPress Plugin Work: What We Know
What Anthropic’s docs say about Claude Opus 5.5, how our team routes plugin work between Opus and Sonnet, what changes for API users, and a ten-ticket test you …
Writing · Business & Agency
Agency life, client stories, team management, and business insights from running a WordPress agency.
What Anthropic’s docs say about Claude Opus 5.5, how our team routes plugin work between Opus and Sonnet, what changes for API users, and a ten-ticket test you …
How a Claude Code skill actually loads, what belongs in SKILL.md versus references, how to write a description that triggers, why oversized skills hurt after co…
We moved four headless WordPress sites from rebuilding on every publish to rendering the blog on request from a Workers KV cache. Why, how it works, what bit us…
Three October 2026 reports showed the same pattern: AI agent tooling leaks credentials sideways. Where secrets escape in an agent setup, the rules we run, and a…
Why we stopped building one integration per store: Learnomy 2.1.2 and BuddyNext 1.2.5 let Make, Zapier and n8n sell courses and membership plans through one sig…
Listora is our first plugin on the EmDash registry. How the port from WordPress went, how publishing works, what broke, and what is still not ready.…
Plugin revenue is down across platforms, not only WordPress. Split new-customer revenue from renewals to see where each product really stands.…
Cloudflare acquires Deno: Deno Deploy closes in six months, Workers stays the same, and self-hosting is promised. What agencies should check and tell clients.…
Two numbers decide whether a WordPress agency is profitable: a $5,000 average project and 25% recurring revenue. The 2026 survey data and a one-hour test.…
Six real Claude Code and Git incidents from a 100+ repo agency, with the hook, setting or habit that stops each one from happening again.…
An MCP server is a dependency that changes without telling you. Six real incidents from our agency, and the pin, test and watch rules we keep now.…
When agents write most of the code, the job becomes deciding, checking and owning. Here is how we design the work so people do not become tired reviewers.…