My comment patrol: one prompt that answers my Threads comments every 6 hours

You post, a few comments roll in, and then you spend the next two hours checking your phone. Every reply you write late is a conversation that dies early. Early replies are the ranking factor that matters most on Threads, and they are the easiest ones to outsource.

The task I gave my MyClawBots agent: run a comment patrol. Every 6 hours it checks the posts from the last 7 days, reads new comments, and replies to the ones worth answering. The run I looked at this week posted exactly one verified reply on a quiet afternoon. That is the point. Nothing worth answering gets missed just because I was busy.

The skill doing the work

This runs on the Threads skill: the agent reads your posts and comments through the MyClawBots social connection and replies through the same pipe. No app-switching, no notification scrolling, no draft-then-copy-paste ritual. One schedule, six-hour cadence, done.

The prompt below is the real one. Paste it into your own MyClawBots agent, adjust the voice line, and you have the same patrol.

Every 6 hours, answer the new comments on my Threads posts.

1. Read my recent Threads posts, then get the comments on posts published in the last 7 days.
2. Skip anything older than 7 days, my own comments, spam, pure self-promo, and hostile comments.
3. Reply to each remaining comment in a short, plainspoken voice. Keep it human. No corporate polish.
4. Send each reply as a reply to the comment (reply_to_id), never as a new post.
5. Never reply to the same comment twice. Keep a running list of comment IDs you already handled and check it before posting.
6. Never retry a reply whose response was ambiguous. If you are unsure it landed, read the comment list first and only then decide.

Report back in one line: how many replies posted, or nothing new worth answering.

What actually broke

Rule 6 exists because we learned it the hard way. The publishing endpoint sometimes returns a garbled or truncated response, which looks exactly like a failure. Our early patrol runs did the "reasonable" thing and retried. The first reply had actually landed, so the retry posted a duplicate, and in one case the same reply went out three times. Public duplicates on a public account. That is the lesson we built the rule from: an ambiguous publish response is never an excuse to re-send. Verify against the comment list first, and treat an inconclusive check as "do not retry," never as "probably failed."

Two more honest limits. First, comments do not always show up in the read API immediately; a reply can take a few minutes to appear, so the verify step needs patience. Second, the agent only replies where you have a connected account. No connected Threads account means no patrol, and one-off publishes still need your tap in the approval flow. If you run scheduled posts through the same agent, those are pre-approved, but ad-hoc work waits for you.

Copy this to start

Take the prompt above, swap in your account, and tighten the voice rule to match how you actually write. Then ask yourself one question after a week: which conversations did the patrol keep alive that you would have let die? Pricing is flat, Lite $19, Pro $39, Max $79, Team $59 per seat, with a 3-day trial that needs no credit card. The bill is the same whether the patrol runs quiet for a month or answers fifty comments a day.