In a previous post we looked at using Claude Cowork for ecommerce—including support ticket analysis: export tickets from Zendesk, Help Scout, or your help desk into a folder, and have Cowork summarize themes, sentiment, and pain points. To make that repeatable and consistent, you can turn it into a skill: a reusable workflow Claude follows whenever you run that type of task. Here’s how to develop a skill for support ticket analysis (use case 5) so you can run it again and again with minimal re-explaining.
What is a Claude skill?
A skill teaches Claude a specific workflow once. It has a name, a description (which helps Claude decide when to use it), and a set of instructions in a SKILL.md file. When you describe a task that matches the skill, Claude can activate it and follow those instructions instead of improvising from scratch. For Cowork, that means: you point Claude at a folder of ticket exports and ask for an analysis; the skill defines exactly how to read, categorize, and summarize so the output is consistent every time.
Step 1: Define what the skill does
Be clear about inputs, steps, and outputs.
- Inputs: A folder containing exported support tickets (CSV, JSON, or text/HTML exports from Zendesk, Help Scout, Intercom, etc.). Optionally a short note (e.g. “focus on last 30 days” or “ignore billing”).
- Operations: Read all ticket files in the folder; extract subject, body, date, and any tags/status; group by theme or category; assess sentiment where useful; summarize recurring issues and pain points; produce a short report or structured summary (e.g. markdown or a spreadsheet).
- Output: A single report or spreadsheet: themes, top issues, sentiment snapshot, and optionally suggested action items (e.g. “add FAQ,” “escalate to product”).
Write this down in one paragraph. That becomes the backbone of your description and instructions.
Step 2: Choose a name and write the description
Use a lowercase, hyphenated name (e.g. support-ticket-analysis). The description is what makes Claude select this skill, so be specific. Example:
“Analyzes exported support tickets from a folder: reads CSV/JSON/text exports from Zendesk, Help Scout, or similar; extracts themes, recurring issues, and sentiment; produces a summary report or spreadsheet with top pain points and optional action items. Use when the user has a folder of support ticket exports and wants a consolidated analysis or trend report.”
That way, when you say “analyze the tickets in this folder” or “run a support ticket report,” Claude knows this skill applies.
Step 3: Create the SKILL.md and instructions
Create a file named SKILL.md. At the top, add YAML front matter with name and description (same as above). Below that, write the instructions in clear, numbered steps. For example:
- List all files in the authorized folder and identify which are ticket exports (by name or extension).
- For each file, read and parse (CSV, JSON, or plain text). Extract at least: ticket id, subject/title, body/description, date, status/category if present.
- Combine all tickets into a single logical set. If the user specified a date range or filter in their message, apply it.
- Group tickets by theme or category (e.g. shipping, returns, login, pricing). Use consistent labels.
- For each theme, note frequency and, where useful, sentiment (e.g. “mostly neutral,” “frustrated,” “urgent”).
- Write a short summary report (markdown or similar) with: overall ticket count and date range; top 5-7 themes with counts and one-line descriptions; 2-3 recommended action items. Optionally produce a simple spreadsheet (e.g. themes, count, sample quotes).
- Save the report (and spreadsheet if any) in the same folder or a clear subfolder, and tell the user where the output is.
Add one or two usage examples (e.g. “User puts 50 Zendesk CSV exports in ~/Downloads/tickets and says ‘Analyze these for last month.’ Output: Support_Analysis_Report.md and optionally Support_Themes.csv.”). Include a note on edge cases: e.g. if no ticket files are found, say so and suggest the user check the folder or export format.
Step 4: Use the skill with Cowork
In Claude Desktop, enable Cowork and grant folder access to the directory where you’ll put ticket exports. Put your skill where Claude can load it (per Anthropic’s current skill setup). Then: drop your ticket exports into the folder, and in chat say something like “Analyze all support tickets in this folder and give me a theme and sentiment report.” Claude should activate the support-ticket-analysis skill and run the steps in SKILL.md. Cowork will read and write files in that folder and produce the report.
Step 5: Test and refine
Run the skill on a small set of exports first. Check that themes make sense, counts are right, and the output format is useful. If your help desk export format is different, add a line in the instructions for that format. Iterate on the description if Claude doesn’t activate the skill when you expect. After a few runs, you’ll have a stable skill you can reuse weekly or monthly for support ticket analysis.
Why this is useful for ecommerce
Support teams in ecommerce deal with returns, shipping, orders, and product questions. A repeatable ticket-analysis skill gives you consistent visibility into what’s trending—without manually tagging or building dashboards. You export, drop into a folder, run the skill in Cowork, and get a report. That feeds into FAQs, product fixes, and training.
Claudery connects your WordPress and WooCommerce store to Claude and Cursor for live store operations. Pair that with Claude Cowork and skills like this for file-based analysis so both your store and your support data stay in the loop.
