by mouryanE
Turns a WhatsApp group export into a prioritised issue backlog, executive brief, and searchable HTML triage board — in one go.
Most support teams manage issues through WhatsApp groups — but nothing in that chat ever becomes a proper task list. Messages get buried, problems repeat, and there's no way to know what's actually open vs fixed.
chatanalys-e reads your WhatsApp export and turns the entire conversation history into a structured, prioritised backlog — the same thing a project manager would produce after reading 4,000 messages, done in minutes.
A 6-person SaaS support team has been running their customer escalations through a WhatsApp group for 18 months. They export the chat and say:
"analyze my WhatsApp support chat"
In minutes they get a ranked backlog of 240 tasks, grouped by module (Billing, Integrations, Mobile App). The triage board shows 12 critical issues still unresolved, 3 of them security-related. The executive brief highlights that the Billing module has the highest recurrence rate — the same payment failure has been reported 34 times.
A fleet operations manager uses a WhatsApp group to coordinate between drivers, mechanics, and dispatchers. After exporting 3 months of chat:
"extract issues from this WhatsApp export — build me a task list"
The skill identifies 87 distinct issues: vehicle breakdowns, document expiry warnings, route complaints. Contract IDs like SABAA21009833CO are automatically linked to every message that mentioned them, so the ops manager can see exactly which vehicle each fault belongs to.
A team member who joined the group 6 months ago discovers their admin exported the full group history going back 2 years. They already ran the analysis on their own 6-month window. They attach the full export and say:
"extract issues from WhatsApp export"
The skill detects the gap and reports:
"Already analyzed: 01 Mar 2026 → 10 Sep 2026. New export covers: 01 Jan 2024 → 10 Sep 2026. Missing: 01 Jan 2024 → 28 Feb 2026."
They choose Process missing only. The historical issues are merged into the existing backlog, with earlier first_raised dates pushing some recurring bugs up the priority ranking.
A team runs the analysis every Friday. They export the chat and say:
"analyze my WhatsApp support chat"
The skill finds 47 new messages since last Friday, processes only those, adds 3 new tasks, and updates occurrence counts on 6 existing ones. The triage board refreshes in place.
- Open the WhatsApp group on your phone
- Tap the group name → More (three dots) → Export Chat
- Choose Without Media (media files are not needed)
- Save or share the
.zipfile — it contains a_chat.txtinside
Attach the .zip or _chat.txt to your Claude session and say any of:
- "analyze my WhatsApp support chat"
- "extract issues from WhatsApp export"
- "turn this WhatsApp chat into a task list"
- "build a backlog from our WhatsApp group"
Claude will take it from there.
If you've run this before, Claude checks what's already been analyzed and shows you exactly what's new:
"Already analyzed: 15 Jun 2025 → 10 Sep 2026 (4,655 messages, 391 tasks) New export covers: 01 Jan 2021 → 10 Sep 2026 Missing: 01 Jan 2021 → 14 Jun 2025 (pre-join history)"
It then asks how you want to proceed:
- Process missing only — analyze the gap and merge into your existing backlog
- Full re-analysis — reprocess everything from scratch
- Show existing results — open the current board without processing anything
Nothing runs until you choose. Nothing gets overwritten unless you say so.
Every message in the export is parsed into a structured dataset: sender, timestamp, message type, text, mentions, and any attachment references.
Messages are classified into themes (Billing, Contracts, Security, Performance, etc.). Contract and entity IDs are extracted with their full mention history.
Messages are grouped into conversation sessions (90-minute gap = new session). Issue-bearing sessions are filtered and split into 10 parallel batches.
10 AI agents run simultaneously, one per batch, each extracting structured issue records with title, module, severity, status, description, suggested fix, and evidence quotes.
All extracted issues are deduplicated and priority-scored. The result is a ranked backlog.
| File | Contents |
|---|---|
output/messages.csv / .json |
Every parsed message with theme classification |
output/references.csv / .json |
All contract/entity IDs and their mention history |
output/themes_by_month.csv |
Theme volumes month by month |
output/analysis.json |
Turnaround times, load by hour/day, top senders |
output/backlog.csv / .json |
Prioritised deduplicated task list (importable to Jira/Linear) |
output/issues_raw.json |
All raw extracted issues before deduplication |
output/backlog_brief.md |
Executive brief — headline findings, security incidents, fix priority list |
output/triage_board.html |
Standalone searchable board (works offline, no server needed) |
| Factor | Points |
|---|---|
| Critical severity | 120 |
| High severity | 70 |
| Security incident | +60 bonus |
| Outage/incident | +40 bonus |
| Unresolved status | +35 |
| Each recurrence (up to 20) | +3 each |
| Open longer than 60 days | +15 |
| Still active (raised in last 6 months) | +20 |
mouryanE
