Transform meeting transcripts into brutally honest, insight-rich dashboards
Takes .vtt transcript files from Microsoft Teams meetings and generates comprehensive multi-tab HTML dashboards with:
- 📋 Executive summaries and action items
- ⏱️ Timeline analysis with time waste identification
- 👥 Participant analysis with honest feedback
- 💡 Hidden insights and elephants in the room
- 📊 Forensic analytics with meeting cost calculations
- 📈 Longitudinal trends and accountability tracking
This isn't a feel-good summarizer. The analysis agents are designed to:
- Find what's NOT being said (avoidance patterns)
- Quantify waste (time and money)
- Expose dysfunction (recurring issues, accountability failures)
- Make predictions (where problems are heading)
- Deliver uncomfortable truths (that people need to hear)
| Agent | Purpose |
|---|---|
timeline-analyzer |
Forensic time analysis, identifies waste and what was avoided |
people-analyzer |
Value-per-minute scoring, power dynamics, honest feedback |
insights-generator |
Elephants in the room, risk radar, buried opportunities |
analytics-generator |
Meeting cost in $, dysfunction metrics, A-F grading |
longitudinal-analyzer |
Accountability audit, recurring issue tracking, predictions |
consolidator |
Harmonizes all outputs with consistent naming |
Transcript (.vtt)
↓
┌──────────────────────────────────────────────────┐
│ PARALLEL ANALYSIS (5 specialized agents) │
└──────────────────────────────────────────────────┘
↓
┌──────────────────────────────────────────────────┐
│ CONSOLIDATION │
│ • Name normalization │
│ • Cross-reference validation │
│ • Insight amplification │
└──────────────────────────────────────────────────┘
↓
┌──────────────────────────────────────────────────┐
│ DASHBOARD GENERATION (SSW branded) │
└──────────────────────────────────────────────────┘
↓
Deploy to Azure Blob Storage (dashboards.sswtiger.com)
MeetingSummary/
├── .claude/
│ ├── agents/ # Analysis agents (Critical Edition)
│ │ ├── timeline-analyzer.md
│ │ ├── people-analyzer.md
│ │ ├── insights-generator.md
│ │ ├── analytics-generator.md
│ │ ├── longitudinal-analyzer.md
│ │ └── consolidator.md
│ └── skills/ # Claude skills (auto-triggered)
│ ├── organize-transcript/
│ ├── analyze-meeting/
│ ├── generate-dashboard/
│ ├── deploy-dashboard/
│ └── list-projects/
├── templates/
│ └── dashboard.html # SSW-branded dashboard template
├── projects/ # .gitignored - contains sensitive data
│ └── {project-name}/
│ ├── transcripts/ # .vtt files
│ ├── analysis/ # Agent outputs (JSON)
│ └── dashboards/ # Generated HTML
├── CLAUDE.md # Claude instructions
└── README.md
- Claude Code CLI
- Node.js
- Azure CLI (for dashboard deployment)
# 1. Clone and install
git clone https://github.com/SSWConsulting/SSW.Tiger.git
cd SSW.Tiger
npm install
# 2. Configure environment
cp .env.example .env
# Edit .env and set:
# CLAUDE_CODE_OAUTH_TOKEN (or ANTHROPIC_API_KEY)
# DASHBOARD_STORAGE_ACCOUNT=<your-storage-account>
# DASHBOARD_BASE_URL=dashboards.sswtiger.com
# COSMOS_ENDPOINT=https://<your-cosmos>.documents.azure.com:443/
# 3. Login to Azure (one-time, for dashboard deployment)
az loginclaudeThen talk naturally:
"Here's the weekly standup for project-alpha" + attach your .vtt file
"Process the yakshaver transcript from today"
Claude generates the dashboard HTML. To deploy it afterwards:
"Deploy the dashboard" # Tell Claude (uses deploy-dashboard skill)
Or deploy manually from the command line:
node processor/deploy-local.js <project-name> <meeting-id>Local Development:
# Build and run
docker-compose build
docker-compose run --rm meeting-processor /app/dropzone/meeting.vtt projectnameProduction (Azure):
See TIGER.md for full Azure deployment guide with Key Vault integration.
The Azure pipeline handles deployment automatically via processor/index.js → deployer.js.
The processor supports two authentication methods:
Best for testing, low volume, variable usage.
export ANTHROPIC_API_KEY=sk-ant-api03-...
# or
export CLAUDE_API_KEY=sk-ant-api03-...Best for production, high volume, predictable usage.
export CLAUDE_SUBSCRIPTION_TOKEN=your-subscription-tokenPriority: If both are set, subscription is used (lower per-request cost).
For production deployments, store credentials in Azure Key Vault:
# Store secrets
az keyvault secret set --vault-name kv-tiger --name claude-api-key --value "..."
az keyvault secret set --vault-name kv-tiger --name claude-subscription-token --value "..."
# Reference in Container App
az containerapp job secret set --name job-tiger-processor --resource-group rg-tiger \
--secrets "claude-api-key=keyvaultref:https://kv-tiger.vault.azure.net/secrets/claude-api-key"See TIGER.md for complete setup instructions.
- Open Teams Calendar → select the meeting
- Go to Recordings & Transcripts tab
- Download as
.vtt - Place in
projects/{project}/transcripts/
Transcripts are stored in:
- Private meetings: OneDrive →
Recordingsfolder - Channel meetings: SharePoint → Team site →
Recordingsfolder
| Tab | Content |
|---|---|
| Overview | Summary, decisions, action items, hard truths |
| Timeline | Segments, time waste analysis, flow analysis |
| People | Participant cards, team dynamics, power dynamics |
| Insights | Team health, elephants, risk radar, notable moments |
| Analytics | Cost analysis, charts, metrics, dysfunction metrics |
| Trends | Trajectory, accountability audit, recurring issues, predictions |
The dashboard follows SSW design guidelines:
- Primary Red:
#CC4141 - Charcoal:
#333333 - Font: Inter (web) / Helvetica Neue (print)
- Logo motif: Four colored squares
- Meeting summary and key decisions
- Action items with owners
- Timeline segments with energy levels
- Time waste quantification (minutes + dollars)
- Participation inequality (Gini coefficient)
- Value-per-minute scoring (not just airtime)
- Power dynamics (interruptions, who talks over whom)
- Elephants in the room (what's being avoided)
- Risk radar (technical, people, process risks)
- Recurring issues (the graveyard of unresolved problems)
- Predictions (where problems are heading)
- Meeting grades (A/B/C/D/F with justification)
Contains instructions for Claude on how to process transcripts and generate dashboards.
After processing a transcript, you'll get:
-
Analysis files in
projects/{project}/analysis/:timeline.jsonpeople.jsoninsights.jsonanalytics.jsonlongitudinal.jsonconsolidated.json← Used for dashboard
-
Dashboard in
projects/{project}/dashboards/{date}/index.html -
Deployment at
https://dashboards.sswtiger.com/{project}/{meeting-id}
The agents are in .claude/agents/. Each agent has:
- A specific analysis focus
- Output format specification
- "Hard truths" philosophy
Feel free to tune the agents for your team's needs!
Internal SSW tool.
Built with Claude | Styled with SSW Brand Guidelines