Agentic coding freed my content from WordPress (mkennedy.codes)

A long-time blogger used agentic AI (Cursor + Anthropic’s Claude Sonnet-4) to migrate 164 posts from an old WordPress blog to a Hugo site in about two hours. The agent fetched content, converted posts to Markdown, copied and relinked binary assets (images and files), and iterated until the output matched the author’s expectations. The migrated posts were organized under a clear archival path (/posts/r/slug), and the author implemented 301 redirects in NGINX so old WordPress URLs like blog.example.net/YYYY/MM/DD/slug reliably map to the new site. The exact redirect used: location ~ "^/([0-9]{4})/([0-9]{2})/([0-9]{2})/(.+?)/?$" { return 301 https://mkennedy.codes/posts/r/$4/; }. This is a practical demonstration of agentic coding delivering real-world value: it saved time and recurring domain costs, preserved SEO through proper redirects, and handled tedious, error-prone tasks end-to-end. For the AI/ML community it underscores how orchestration of LLMs and developer tooling can automate complex migration workflows, not just generate code snippets. Key technical takeaways: ensure iterative validation of converted content and asset links, verify regex redirects against edge cases, and keep human oversight for SEO and URL canonicalization. The story highlights both the power and the workflow considerations when trusting AI agents with production migrations.