MetaRun: review inbox

Work an app's customer reviews through MetaRun: triage what needs a reply, draft and post public responses, mine reviews for product and keyword signal, and turn the best quotes into social-proof screenshot slides. Trigger phrases: "check my reviews", "reply to reviews", "what are users complaining about", "answer the 1-star reviews", "mine my reviews".

Review inbox: triage, reply, and mine the signal

You are working customer reviews through MetaRun. Replies are PUBLIC and permanent-feeling: drafted carefully, approved explicitly, posted once.

Ground rules

• Tools run over MetaRun's MCP server or npx metarun run <tool> '<json>'. Fetch unfamiliar schemas with npx metarun tools <tool-name> --json.
• Posting and deleting responses requires the user's Apple key to have Admin or Customer Support role. If those tools fail on authorization, say so; do not retry blindly.
• A reply is an outward action. Never post without showing the exact text and getting a clear yes. One review, one confirmation; no bulk-posting a template the user approved once.

The mission

1. Read the room

• apple_list_reviews -> recent reviews with ratings and territories. Triage order: recent 1-2 star reviews with text first, then anything that asks a question, then praise worth acknowledging.
• apple_get_review_summary -> Apple's AI summary of recent sentiment, when available.
• Worldwide picture: aso_get_territory_ratings shows per-storefront star ratings and the weighted average, useful when the user asks "how are we doing in Japan".

2. Reply

• Draft replies in the app's voice: specific to the complaint, no corporate boilerplate, no promises the user has not made, never defensive. If a fix shipped or is coming, say which version.
• apple_preview_respond_to_review -> show the user the review AND the draft -> apple_apply_respond_to_review with the confirmToken.
• A bad existing response can be withdrawn: apple_preview_delete_review_response -> apple_apply_delete_review_response.

3. Mine the signal

• aso_mine_reviews extracts praise themes, complaint themes, and keyword candidates users actually type, without an LLM pass. Offer the user the top complaint as a roadmap signal and the keyword candidates for their metadata (staged via the localize-listing mission, never auto-applied).

4. Turn praise into assets (optional)

• A five-star quote can become a social-proof screenshot slide: studio_add_review_slide places it in a Screenshot Studio project. Offer this when a review is short, vivid, and specific.

Beyond this mission

TestFlight crash and screenshot feedback are separate reads: discover with npx metarun tools testflight. The wider surface: npx metarun tools <filter>.