Weekly SEO grading digest for Contentful blog posts

Every Monday at 9am, pull the last 30 days of published articles from Contentful using the CDA - Get Published Entries operation against our blog content type. Use the Content Delivery API (not the Content Management API) so unpublished drafts are never graded. Only grade the blog content type by default and keep the content type ID configurable so it can point at a different content type later.

For each returned post, grade it on SEO health across five dimensions: title length (aim for 50 to 60 characters), meta description quality (length, specificity, and whether it earns a click), presence of a clear H2 structure in the body, keyword density on the target keyword, and readability. Produce a numeric score per dimension plus a short reasoning note.

Rank all graded posts weakest to strongest and pick the ten weakest. For each of those ten, draft a concrete fix: a stronger rewritten title, a rewritten meta description strictly under 160 characters, and one suggested internal link to another published post in the space that would strengthen topical authority. The fixes should be specific to the post, not generic templates.

Format the ranked list plus fixes as an HTML email digest with each post as its own card showing the current title, current meta description, the scores, and the suggested rewrites side by side with the originals. Send it to the marketing lead using Gmail's Send a Message operation. Keep the subject line dated so weekly digests are easy to scan in the inbox.

The grading and rewriting is judgement-heavy and differs per entry, so run this as an agent rather than a fixed pipeline. Contentful's own product marketing highlights AI-generated meta descriptions and SEO copy as a top use case (see https://www.contentful.com/blog/ai-actions-seo/), and this workflow reproduces that idea as a scheduled digest.