Prompt Recipes for AI Working With WordPress

A good WordPress instruction is more than a clever sentence. It defines the task, the data source, the allowed action, the forbidden action, the expected output and the point at which the assistant must stop. The prompt cannot replace a real permission boundary, but it can make the intent and review process explicit.

This library treats prompts as operating recipes. Every recipe belongs to a task, a level of access and a verification procedure.

The six elements of a reliable instruction

Name the subject, scope, evidence sources, allowed operations, forbidden operations and output format. Add a stopping condition and require the assistant to separate observed facts from recommendations. When content may change, identify the exact records and status that must remain unchanged.

Prompts are not permissions

“Do not publish” is useful intent, but it is not a technical guarantee. Pair the instruction with an identity that lacks publication capability. The strongest demonstration is a prompt that asks the assistant to prepare a draft while WordPress refuses an unapproved publication attempt.

Make the result easy to verify

Ask for stable record identifiers, URLs, current values, proposed values and reasons. Use a table or structured list. Require a dry-run plan before execution. For a large site, begin with a small sample and compare the reported count with WordPress.

Keep secrets and private data outside the prompt

Credentials belong in the client or connector configuration, not in the conversation. Remove personal data from examples and exports. A troubleshooting prompt should ask for sanitized errors and component versions, never raw authorization headers.

A simple way to get useful results

Get structured site information and inspect selected published pages after connecting. No temporary task is needed for this public reading. You can also browse public pages without the plugin; Agent Control adds structured access and a path toward authorized WordPress work. Authorize a draft task and select any reference content. The assistant can create and revise drafts created by that task. Existing references remain read-only, even when a reference is itself a draft. Review the result in WordPress. Approving a content change does not authorize publication. Solo, Pro or Agency must also have a publication task that covers the still-valid approval. Check the published result yourself.

Continue

Start with How to Write Good Instructions for an AI Working in WordPress, then select a task-specific recipe. Use Which WordPress Access Level Should You Give an AI? to make the technical boundary match the written instruction.