See which posts actually draw discussion

Every content plan runs on a hunch about what your readers care about. Traffic numbers tell you what people clicked, which is not the same thing. Comments tell you what people cared enough to type a reply about, and that signal is sitting in your database already, spread across hundreds of individual posts where nobody ever looks at it as a whole.

This prompt gathers it up. It reads a set of posts you name, reads the approved comments on each, and reports where the actual conversation lives. It changes nothing and moderates nothing.

Show me which of my posts actually draw discussion. Read only, change nothing,
and do not moderate anything.

Start by listing [my last thirty published posts] with their titles and
publication dates. Then read the approved comments on each of those posts in
turn.

Rank the posts by how much genuine discussion each one drew, and for the top
few, tell me what the conversation is actually about: questions people keep
asking, disagreements, requests for something you do not cover yet. Then show
me the posts that drew nothing at all.

Be explicit about the sample. Tell me how many posts you covered and how many
comments you read across them, so I know the ranking is about those posts and
not my whole archive. Finish with what this suggests I should write more of.

Why this is safe to run

Both abilities are reads. Listing posts and reading a post’s approved comments report what is already published and cannot approve, edit, trash, or reply to anything. The moderation ability that could act on a comment is a separate switch this recipe never asks for, so a read-only agent connection runs this fine.

Two honest notes on the method, because how the read works decides how far the answer reaches. There is no site-wide comment feed to pull: the comment read works one post at a time, so the agent walks the list of posts you gave it and the ranking covers exactly that list. Name your last thirty posts and you learn about those thirty. It is a survey the agent assembles post by post, not a leaderboard you could look up. And it reads approved comments only, which means a post drowning in unmoderated spam reads as quiet here. If your queue is backed up, clear it first with the triage the comment queue recipe and run this after, or the picture will be wrong in a way that looks perfectly reasonable.

Privacy needs no configuring here. Comment reads never hand back an email address or an IP, so the report is built from author names, content, and timing, and nothing personal leaks into your chat window. Every call runs against the capability of the user you bound the agent to and lands in the Activity Log.

Already connected? Skip the paste. Ask your agent for the “see which posts actually draw discussion” recipe, or just ask which of your recent posts got people talking, and it runs the same reads against your own site.