review-responder
Draft a public reply for every review in a CSV (author, stars, text): 4-5 star reviews get a thank-you that quotes something specific from the review plus an invite back; 1-3 star reviews get empathy and an offline contact path, never defensiveness. Use when answering Google/Facebook reviews instead of staring at a blank reply box.
$ skillbank install review-responder@1.0.0 # pins exactly 1.0.0 · README + changelog included
Review Responder
Overview
Every review deserves a reply, and every reply is public marketing. Review Responder reads a CSV of reviews (author, stars, text) and drafts one reply per review with the tone rules baked in:
- 4–5 stars: thank them by first name, quote something specific they
- 1–3 stars: empathize, apologize for the experience, and take the
said (so it never sounds canned), and invite them back.
conversation offline with a contact path. Never defensive, never argue facts in public — the audience is future customers, not the reviewer.
Drafts are a starting point, not a post button: read each one before it goes live. All data stays local — nothing is uploaded.
Installation
1. Copy the review-responder/ directory anywhere convenient. 2. Make the script executable (Unix):
chmod +x scripts/drafts.py
No pip install needed — standard library only.
Configuration
No config file. Pass your business name and offline contact path as flags:
| Flag | Default | Description | |---------------|---------|------------------------------------------------------| | --business | — | Business name, used in positive replies. | | --contact | — | Phone/email quoted in negative replies ("reach out | | | | to us directly at … so we can make this right"). |
Set --contact to the number or inbox where you actually answer — a dead contact path is worse than none.
Inputs
| Command | Arguments | Returns | |------------------------------------------------------------|---------------------------|----------------------------------| | drafts.py <reviews.csv> [--business NAME] [--contact C] | CSV with author, stars, text | One draft reply per review. |
stars must be 1–5; missing columns, blank authors, or a missing file exit with a clear error and a nonzero exit code — never a raw traceback.
Worked example
$ cat > reviews.csv << 'EOF'
author,stars,text
Maya Thompson,5,"The technician arrived on time and fixed our AC in under an hour. Best service we've had in years!"
Dana Lee,2,"Waited 45 minutes past my appointment and nobody explained why. Frustrating afternoon."
EOF
$ ./scripts/drafts.py reviews.csv --business "Cooper's Auto" --contact "(555) 010-2030"
--- Reply to Maya Thompson (5/5) ---
Thank you, Maya! "The technician arrived on time and fixed our AC in under
an hour." That's exactly the kind of visit we aim for. We'd love to see you
again at Cooper's Auto.
--- Reply to Dana Lee (2/5) ---
Dana, we're sorry your experience wasn't what it should have been. We take
feedback like this seriously and we'd like the chance to fix it. Please
reach out to us directly at (555) 010-2030 so we can make this right.
Safety notes
Never post the drafts unreviewed. If a negative review makes a factual claim you dispute, do not litigate it in the reply — keep the public response short and empathetic and resolve the facts privately. One defensive reply costs more future customers than the bad review did.
Version history
Changelog
1.0.0 — 2026-09-27
- Initial release.
scripts/drafts.pydrafts one public reply per review
from a CSV: specific thank-yous for 4-5 stars, empathetic offline redirects for 1-3 stars. Added references/tone-rules.md.