Guide · Agent skill
How to use the Sentiment Analysis for User Feedback agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The sentiment-analysis skill is a set of written instructions that tells an AI agent how to read a large pile of user feedback, such as reviews, survey answers, CSV files, PDFs or social listening reports, and turn it into a structured first-pass synthesis. This guide is for product managers, researchers and marketers who have more feedback than they can read by hand. By the end you will know how to install the skill, how to phrase a request that triggers it, and what to check before trusting the output.
Sentiment Analysis for User Feedbackphuryn/pm-skills · MIT Download the skill Quick start
- 1Download the ZIP. Get the sentiment-analysis ZIP from this page and unzip it to reveal SKILL.md.
- 2Choose an install location. Place the skill in ~/.claude/skills/ for personal use or in .claude/skills/ inside a project. Alternatively, upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Prepare your feedback. Gather reviews, survey answers, CSVs or PDFs and remove identifiable personal details.
- 4Make a natural request. Describe the data and goal, for example asking for user segments with sentiment scores from a named file.
- 5Review the segment report. Check the segment profiles, quotes and sample sizes against the raw feedback.
- 6Turn findings into decisions. Use the quick wins and prioritized recommendations as a starting point, then validate the most important ones with further research.
What does the sentiment-analysis skill make the agent do?
Once installed, the agent treats your feedback as research material. It identifies at least three user segments, assigns each segment an overall sentiment score from -1 to +1, and summarizes the core job each group is trying to accomplish (jobs-to-be-done).
For each segment, the output is organized into a profile, a sentiment score with satisfaction drivers and detractors, positive themes, pain points with direct quotes where available, a product-segment fit assessment, and two or three prioritized recommendations.
- Segment profiles with common characteristics and approximate share of the dataset
- Jobs-to-be-done and desired outcomes per segment
- Sentiment score from -1 to +1, plus an NPS proxy if applicable
- Strengths, complaints and unmet needs, with quotes
- Quick wins versus strategic initiatives, and which segments to prioritize
How do you install the sentiment-analysis skill?
Download the ZIP file from this page and unzip it. The skill is a single file named SKILL.md. Place the unzipped folder in one of these locations, depending on whether you want it for all projects or one project:
- Personal skills folder: ~/.claude/skills/
- Project skills folder: .claude/skills/ inside your project directory
- Claude apps: upload the skill under Settings → Capabilities → Skills
How do you trigger the skill with a natural request?
You do not need a special command. The skill's description tells the agent when to use it: when you are analyzing user feedback at scale, running sentiment analysis on reviews or surveys, or identifying satisfaction patterns. Name the data and the goal in plain language, and point the agent at the files or text you want analyzed.
The skill expects the feedback sources to be available to the agent. If you have CSV files, PDFs, survey exports or review text, provide them in the conversation or in the working project.
- "Analyze these 400 App Store reviews and group users into segments with sentiment scores."
- "Run sentiment analysis on the attached survey export and tell me which satisfaction patterns stand out."
- "Using the CSV in this folder, identify user segments and their jobs-to-be-done."
What output should you expect?
The skill asks the agent to work through six steps: data ingestion, segment identification, thematic analysis, sentiment scoring, impact assessment and synthesis. The result is a segment-by-segment report rather than a single overall score.
Treat the output as a structured starting point. The scores come from the model reading the text, not from a validated algorithm, so check the quotes and the sample sizes behind each segment before you rely on them.
What do you need before you start?
The source lists no required tools, packages or accounts. The skill does not run scripts and does not use the network. It is only instructions, so the agent does the reading and reasoning itself using the feedback you supply.
The practical limit is the amount of feedback the agent can hold in its context at one time. Very large datasets may need to be split into batches.
What are the safety notes?
Your feedback is pasted or uploaded into the agent's context. The skill gives no privacy guidance, so remove identifiable customer details and avoid sensitive data such as health, financial or identifiable customer records before you start.
- Strip names, emails, account IDs and other identifiers from the files first
- Do not use the skill on health or financial customer records
- Check the agent's findings against the raw data before sharing them
What tips improve the results?
The skill's own best practices favor grounded, balanced output. You can ask the agent to follow them explicitly.
- Ask for quotes and their sources so every finding can be traced back to feedback
- Ask the agent to flag segments with small sample sizes or uncertain sentiment
- Ask for both majority and minority views within each segment
- Ask it to separate feature requests from fundamental pain points
- Ask for cross-segment patterns to find universal pain points
What people use it for
- Summarizing a large batch of app store or product reviews into named user segments
- Building segment profiles for a product or market research report
- Finding satisfaction patterns across customer groups in a survey export
- Prioritizing product improvements from recurring complaints and unmet needs
- Preparing a first-pass synthesis of social listening reports before deeper interviews
Frequently asked questions
Is the sentiment-analysis skill free?
The source lists the skill under the MIT licence. Your agent platform may have its own costs, which the source does not cover.
What do I need to install the sentiment-analysis skill?
The source lists no requirements. Download the ZIP, unzip it, and place it in ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills in the Claude apps.
Does the sentiment-analysis skill run scripts or use the internet?
No. The skill runs no scripts and uses no network. It consists only of a SKILL.md file with written instructions for the agent.
How accurate are the sentiment scores?
The scores are produced by the model reading the text, not by validated algorithms. Small or unrepresentative samples can give misleading numbers, so check them against the raw feedback.
Can I use it for sensitive customer data?
The source warns against it. The feedback is shared with the agent's context and the skill gives no privacy guidance, so avoid health, financial or identifiable customer records.
What should I use instead for rigorous sentiment measurement?
The source does not name alternatives. For statistically rigorous measurement, you would need validated analysis methods and tools outside this skill.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits product managers, researchers and marketers who have a large set of reviews or survey responses and want a structured first-pass synthesis they can check. It is a poor fit if you need statistically validated sentiment numbers, or if your data contains sensitive personal information.
Download the skill