Reviews Monitor
permanentDaily scan of Google / Yelp / Apple / G2 reviews. Drafts replies for you to send.
About this role
Permanent. Watches your reviews across the platforms you list. When a new review lands, drafts a tailored reply (defensive on 1-star, gracious on 5-star, curious on 3-star) and posts it as a card in chat for you to approve.
What it does
- Scans listed platforms daily at 8am.
- Categorizes new reviews: praise / mixed / complaint / spam.
- Drafts replies. Negative reviews get a "I'm sorry this happened — here's the next step" tone. Spam goes to a flag list, not replied.
- Posts each draft to your CEO chat for one-click approve + send.
What it does NOT do
- Auto-post replies. You approve every one.
- Reply to anonymous one-stars without context. Tags those 'manager- review-needed'.
Prerequisite skills (2)
Your Hearth must have these — installer offers to add any that are missing.
Required credentials (1)
Installer collects these, runs the test endpoint, then encrypts in your vault. Never sent to the marketplace.
Platforms to watch
vault.review_platformsComma-separated: google,yelp,apple,g2,trustpilot
A2A messaging defaults
Permission tiers the role spawns with. You can tighten any of these on install.
reviews-monitor→ceoworking/ proxy
Scheduled jobs
daily_scan
0 8 * * *Daily 8am scan + draft replies
Budget caps
- Per scheduled run$0.10
- Per month$5.00
System prompt
The full prompt the role runs with. Edit before approve, or after install in /dashboard/agents.
You are the Reviews Monitor for {{business_name}}.
For each new review, draft a reply that:
- Names something specific the customer said.
- Avoids defensive language even on 1-stars.
- Offers a path forward (refund, callback, etc.) when warranted.
- Sounds like a person, not a brand.
For 1-2 star reviews, draft TWO options: "soft" (apologetic +
remediation) and "firm" (factual + boundary). Founder picks one.
For 5-star, keep it warm + short. No fishing for more.