Building with Claude, prompt patterns, evals, and the engineering reality behind every LLM headline.
The Agentic Browsing audit is now live in Google PageSpeed Insights, no Chrome Canary needed. Paste your URL, read the fractional score, and pass every check that says whether AI agents can read and act on your site.
The five Claude Code tools that change frontend design work: impeccable skill, Figma MCP, claude-in-chrome MCP for visual QA, banana skill for AI image generation, and shadcn/ui MCP.
The five Claude Code tools that change how SEO and content teams work: seo-content skill, seo-technical skill, Firecrawl MCP, Exa MCP, and the filesystem MCP for direct CMS access.
The agency-ops playbook for AI: proposals, SOWs, onboarding documents, SOP creation, meeting summaries, status reports, internal documentation. Where the per-hour gain is highest, and the rules that keep client trust intact.
The five Model Context Protocol servers worth installing today: filesystem, GitHub, Postgres, claude-in-chrome (browser), and Sentry. With install commands, the tools they expose, and the security model.
Five Claude Code skills that turn it from a chat tool into a development environment: impeccable for UI design, playwright-cli for browser automation, gsd for project planning, claude-seo for content and SEO, and banana for image generation.
The content-role playbook for AI on WordPress sites: outlines and briefs, article refreshes, title generation, format conversion, fact-check passes, content QA at scale. Plus the strict rules on what stays human-written end to end.
The SEO-role playbook for AI on WordPress sites: internal link audits, schema generation, content-gap analysis, redirect audits, title/meta rewrites, technical SEO triage, and AI-search citability. Plus what stays human.
The sysadmin-role playbook for AI on WordPress infrastructure: log triage, deploy verification, config audits, backup checks, security sweeps, fail2ban rule generation, and the strict read-only discipline that keeps production safe.
Write LLM evals that catch real regressions: pick the right metrics (exact match, LLM-as-judge, embedding similarity), build a golden dataset, run on every PR, and watch the trend over time.
The developer-role playbook for AI on WordPress projects: ACF scaffolding, plugin/theme debugging, WP-CLI orchestration, code review, migration scripts, Figma-to-component, and the senior-review discipline that keeps quality high.
AI for a WordPress agency is not just developer tooling. The role-by-role map: developers, sysadmins, SEO, content, agency ops, business ops. Concrete tooling and prompts per role, what to never delegate.
Build an LLM agent with tool use: the agentic loop, the tool-call format on Anthropic, OpenAI, and Gemini, runnable code in JavaScript and Python, plus the common failure modes.
AI plus SSH plus the standard server toolkit (systemd, nginx, fail2ban, certbot) accelerates infrastructure work without replacing the sysadmin judgment. Log triage, config audits, deploy verification, security review. Plus what to never delegate.
AI for a WordPress agency in 2026: what to actually adopt first, what to skip, the budget realities, and the four operational shifts that turn AI tooling from a curiosity into a load-bearing part of the business.
Add semantic search to an existing MySQL app with MySQL 9's VECTOR type, an embedding model (Voyage, OpenAI), and application-side cosine ranking. No separate vector database needed.
AI plus WP-CLI plus ACF is the canonical pattern for bulk content updates that used to take a careful afternoon. Schema-aware update_field calls, content rewrites at scale, image alt backfills, and the safety patterns that prevent disasters.
The Figma to ACF Flexible Content pipeline used to be the slow part of every agency build. With AI in the loop, you describe the design intent, hand over the Figma JSON or screenshots, and get back working ACF registration plus template partials. Here is the realistic workflow.
How I use Claude CLI to run WordPress and ACF work end-to-end: ACF field group generation, WP-CLI orchestration, log triage, plugin debugging, bulk content ops. Concrete prompts, what it gets wrong, and where it fits in an agency workflow.
Pick the right Claude tier for the job: Haiku for high-volume cheap, Sonnet for the default workhorse, Opus for hard reasoning. With cost math, latency, and a decision matrix.
The WP-CLI patterns that compose well with AI assistants: multi-step plans with checkpoint approval, generated one-off scripts, database surgery, content migrations at scale, and what to never delegate.
Cut LLM API costs by 90% with prompt caching on Anthropic, OpenAI, and Gemini. The math, what to cache, the TTL trade-offs, and code in JavaScript and Python.
Build RAG with embeddings, chunking, and vector search. Covers OpenAI / Voyage / Cohere embeddings, chunk sizes, hybrid search, reranking, and when long context replaces RAG entirely.
Run a local LLM with Ollama: install, pull a model, the hardware floor, picking between Llama, Mistral, and Qwen, and when local is faster than cloud (and when it isn't).
Get reliable JSON out of an LLM with native structured-output modes (Anthropic tool use, OpenAI Structured Outputs, Gemini schema), plus Zod / Pydantic validation as a fallback.
Six techniques that actually reduce LLM hallucination: grounding with retrieved context, citation requirements, tool use for facts, structured outputs, explicit don't-know permission, and LLM-as-judge verification.
Write an effective system prompt for an LLM with five parts: role, capabilities, constraints, output format, refusal policy. With before/after examples and the structure that maximises cache hits.