From digital-marketing
Generate professional, empathetic, on-brand responses to online customer reviews. Analyzes sentiment, detects severity, adapts tone, and provides operational suggestions. Supports hospitality (Airbnb, Booking, Tripadvisor) and e-commerce/app (Amazon, App Store, Trustpilot) with sector-specific patterns. TRIGGER WHEN: review, recensione, reply to review, respond to review, risposta recensione, customer review, negative review, bad review, rispondere alla recensione, gestione recensioni, review response. DO NOT TRIGGER WHEN: the task is outside the specific scope of this component.
npx claudepluginhub acaprino/alfio-claude-plugins --plugin digital-marketingThis skill uses the workspace's default tool permissions.
Generate a professional, empathetic response to a customer review. Analyze the review, craft an adaptive response, and provide operational suggestions.
Guides Payload CMS config (payload.config.ts), collections, fields, hooks, access control, APIs. Debugs validation errors, security, relationships, queries, transactions, hook behavior.
Builds scalable data pipelines, modern data warehouses, and real-time streaming architectures using Spark, dbt, Airflow, Kafka, and cloud platforms like Snowflake, BigQuery.
Builds production Apache Airflow DAGs with best practices for operators, sensors, testing, and deployment. For data pipelines, workflow orchestration, and batch job scheduling.
Generate a professional, empathetic response to a customer review. Analyze the review, craft an adaptive response, and provide operational suggestions.
The user pastes a customer review (or invokes via /reply-to-customer-review). Optional parameters:
If invoked without arguments, prompt the user to paste a review.
Execute all three steps inline. Do NOT spawn subagents.
Analyze the review and determine:
Language -- identify the review language. For mixed-language reviews, identify the dominant language and note secondary languages.
Sentiment -- classify as one of:
Key Points -- extract the specific topics mentioned (e.g., cleanliness, shipping speed, product quality, staff attitude, app stability, price, location).
Severity (negative/mixed reviews only) -- assess as one of:
Sector -- auto-detect from vocabulary and context:
Load the appropriate reference file: references/hospitality-patterns.md or references/ecommerce-patterns.md. If generic, use general best practices.
Craft the response following these rules:
Tone: Use the user-specified tone or default to professional-empathetic. The tone scale:
formal -- corporate, third-person, measured languagefriendly -- warm, first-person, conversational but professional (this is the professional-empathetic default)casual -- relaxed, direct, uses contractions and informal phrasingLanguage: Respond in the same language as the review unless --lang overrides.
Brand: If --brand is provided, sign off with the brand name. If not, use a generic professional sign-off.
Adaptive strategy for negative reviews:
| Severity | Strategy | Key Elements |
|---|---|---|
| UNFOUNDED | Diplomatic-defensive | Acknowledge feelings, provide factual context, invite private contact |
| MINOR | Empathetic-proactive | Thank for feedback, acknowledge the issue, describe corrective action taken, invite return |
| MAJOR | Empathetic-proactive (urgent) | Sincere apology, take full responsibility, describe immediate corrective action, offer compensation, provide direct contact for follow-up |
| ABUSIVE | Minimal/No response | If responding: brief, professional, offer private channel. May recommend not responding and reporting to platform instead |
For positive reviews: Thank sincerely, reference specific points mentioned, reinforce the positive experience, invite return/continued use. Keep it genuine -- avoid sounding templated.
For neutral reviews: Thank for taking the time, address any suggestions, invite further engagement.
Response length guidelines:
Before presenting the response, self-check against these rules. The response must sound like a real person wrote it -- not a chatbot or corporate template.
Avoiding AI patterns is half the job. Soulless "clean" writing is just as obvious.
"testament", "pivotal", "landscape", "delve", "foster", "underscore", "showcase", "showcases", "vibrant", "crucial", "enhance", "garner", "interplay", "tapestry", "endeavor", "embark", "paramount", "comprehensive", "furthermore", "henceforth", "additionally", "noteworthy", "commendable", "invaluable", "exceptional"
Content: no inflated significance ("This truly reflects..."), no promotional tone (reply as a person, not a brochure), no vague attributions ("many guests say..."), no formulaic "challenges and future" framing.
Language: no copula avoidance ("serves as" -- just use "is"), no -ing fillers at sentence start ("Highlighting..."), no negative parallelisms ("not only... but also"), no forced rule of three, no elegant variation / synonym cycling (pick one word and reuse it -- don't rotate "stay/sojourn/visit/experience"), no false ranges.
Style: no excessive em dashes, no bold text or emoji, no curly quotation marks.
Communication: no servile tone ("Great feedback!", "Thank you so much for your kind words!", "We truly appreciate..."), no template phrases ("We are sorry to hear...", "Thank you for taking the time...", "Your feedback is valuable...", "Your satisfaction is our priority..."), no filler phrases ("In order to", "At its core", "It is important to note"), no generic positive conclusions ("we look forward to continuing", "we strive for excellence"), no excessive hedging ("perhaps", "we believe", "we hope" -- be direct), no collaborative artifacts ("Hope this helps!", "Don't hesitate to...").
After drafting, re-read the response and ask: "Would a real B&B owner actually write this?" If any sentence sounds like ChatGPT, rewrite it.
Present three clearly separated sections:
RESPONSE
The ready-to-copy response text (after AI trace removal), formatted for the review platform. No markdown formatting -- plain text that can be pasted directly.
ANALYSIS
| Field | Value |
|---|---|
| Language | [detected language] |
| Sentiment | [POSITIVE/NEUTRAL/NEGATIVE/MIXED] |
| Severity | [UNFOUNDED/MINOR/MAJOR/ABUSIVE or N/A] |
| Sector | [HOSPITALITY/ECOMMERCE/GENERIC] |
| Key Points | [comma-separated list] |
| Flags | [any concerns requiring attention, or "None"] |
OPERATIONAL SUGGESTIONS
Bulleted list of recommended internal actions based on the review content. Examples:
If the review is positive with no issues, suggest ways to leverage it (e.g., "Consider featuring this review on your website", "Share with the team as positive feedback").
After presenting the output, the user may request adjustments:
Regenerate only the RESPONSE section when adjusting. Keep Analysis and Operational Suggestions unchanged unless the user specifically asks to revise them.