Dokdo Telemetry MCP Server

Collects, scrubs, and reports anonymized usage analytics from Dokdo customer OpenClaw instances.

Overview

This MCP server runs as a sidecar on each customer's OpenClaw VM. It:

1. Reads OpenClaw session logs (JSONL format)
2. Parses conversation events (messages, calls, errors)
3. Categorizes events by topic, language, outcome
4. Scrubs all PII (customer names, numbers, message content)
5. Encrypts the batch
6. Submits daily to Dokdo's central telemetry server

What Gets Collected

Message Events

• Timestamp
• Language detected (English, Vietnamese, Korean, Spanish)
• Topic category (appointment, hours, menu, review, etc.)
• AI confidence score
• Outcome (answered, escalated, error)
• Message length (not content)
• Whether it had attachments

Call Events

• Duration (seconds)
• Language
• Topic category
• Whether call was transferred to human
• Outcome

Integration Health

• API calls to integrations (Clover, Square, Google, etc.)
• Success/failure rate
• Average latency per provider

Toast POS Events (if configured)

• Menu snapshots (items, categories, modifiers)
• Order events (timestamp, status, totals, lineItems)
• Inventory snapshots (items, quantities, categories)
• Labor time entries (shifts, clock in/out)
• Table snapshots (availability, status)
• All PII is scrubbed: customer names, payment info, delivery addresses removed

System Metrics

• Uptime
• Memory/CPU usage
• Timestamp

What Does NOT Get Collected

❌ Customer names, phone numbers, emails (even in Toast orders)
❌ Message content or call transcripts
❌ Payment info or card data
❌ Delivery addresses or customer contact info
❌ Personal health information
❌ Anything PII-adjacent

Installation

1. Deploy to Customer VM

# Clone / download the MCP
git clone https://github.com/dokdosolutions-us/dokdo-telemetry-mcp.git
cd dokdo-telemetry-mcp

# Install dependencies
npm install

# Build
npm run build

2. Configure Environment

cp .env.example .env

Edit .env with customer details:

DOKDO_CLIENT_ID=pho-restaurant-001
DOKDO_API_KEY=sk_dokdo_xxxxx
OPENCLAW_SESSIONS_PATH=/home/username/.openclaw/agents/main/sessions

3. Run as Service

Option A: Systemd (Linux)

sudo cp dokdo-telemetry.service /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl enable dokdo-telemetry
sudo systemctl start dokdo-telemetry
sudo systemctl status dokdo-telemetry

dokdo-telemetry.service:

[Unit]
Description=Dokdo Telemetry MCP Server
After=network.target

[Service]
Type=simple
User=openclaw
WorkingDirectory=/opt/dokdo-telemetry-mcp
ExecStart=/usr/bin/node dist/index.js
Restart=always
RestartSec=10
StandardOutput=append:/var/log/dokdo-telemetry.log
StandardError=append:/var/log/dokdo-telemetry.log

[Install]
WantedBy=multi-user.target

Option B: Docker

FROM node:20-alpine
WORKDIR /app
COPY . .
RUN npm install && npm run build
CMD ["node", "dist/index.js"]

docker run -d \
  --name dokdo-telemetry \
  -e DOKDO_CLIENT_ID=pho-restaurant-001 \
  -e DOKDO_API_KEY=sk_dokdo_xxxxx \
  -v /home/user/.openclaw:/home/user/.openclaw:ro \
  dokdo-telemetry:latest

Option C: Cron Job (Simple, manual daily run)

# Add to crontab (daily at 11:59 PM)
59 23 * * * cd /opt/dokdo-telemetry-mcp && /usr/bin/node dist/index.js >> /var/log/dokdo-telemetry.log 2>&1

4. Verify

# Check logs
tail -f /var/log/dokdo-telemetry.log

# Should see:
# [Dokdo Telemetry] Initialized for client: pho-restaurant-001
# [Dokdo Telemetry] Starting collection run at ...
# [Dokdo Telemetry] Found X events
# [Dokdo Telemetry] Submitted X events for YYYY-MM-DD

Data Format

Daily Batch Submitted to Dokdo Server

{
  "clientId": "pho-restaurant-001",
  "date": "2026-06-06",
  "totalEvents": 147,
  "events": [
    {
      "type": "message",
      "timestamp": "2026-06-06T14:32:00Z",
      "language": "vietnamese",
      "topic": "appointment_inquiry",
      "confidence": 0.96,
      "outcome": "booking_created",
      "messageLength": 42,
      "hasAttachment": false
    },
    {
      "type": "call",
      "timestamp": "2026-06-06T15:12:00Z",
      "duration": 180,
      "language": "english",
      "topic": "hours_inquiry",
      "handoff": false,
      "outcome": "answered"
    }
  ],
  "integrationHealth": {
    "clover_api": {
      "calls": 145,
      "failures": 0,
      "avgLatency": 0.82
    },
    "google_api": {
      "calls": 23,
      "failures": 1,
      "avgLatency": 1.24
    }
  },
  "systemMetrics": {
    "timestamp": "2026-06-06T23:59:00Z",
    "uptime": 2592000,
    "memoryUsage": { "heapUsed": 45000000 },
    "cpuUsage": { "user": 1000000, "system": 500000 }
  }
}

Encryption

Batches are encrypted with AES-256-CBC before transmission. The Dokdo server decrypts using the API key.

Privacy & Compliance

• No PII stored or transmitted: Names, numbers, content all removed
• Encrypted in transit: TLS + AES-256 encryption
• Encrypted at rest: Dokdo server stores encrypted backups
• Data retention: 90 days default (configurable per customer contract)
• Opt-out: Customer can disable telemetry by setting DOKDO_TELEMETRY_ENABLED=false

Troubleshooting

MCP not submitting data

# Check connectivity
curl -I https://telemetry.dokdosolutions.us/ingest

# Verify API key
echo $DOKDO_API_KEY

# Check logs
journalctl -u dokdo-telemetry -f

No events found

# Verify session logs path exists
ls -la /home/username/.openclaw/agents/main/sessions/

# Check file permissions
ls -la /home/username/.openclaw/agents/main/sessions/*.jsonl | head -5

Permission denied

# MCP process needs read access to OpenClaw sessions
sudo usermod -a -G openclaw dokdo-telemetry

Support

• Issues: https://github.com/dokdosolutions-us/dokdo-telemetry-mcp/issues
• Email: [email protected]
• Status: https://status.dokdosolutions.us

Version: 1.0.0
Last Updated: 2026-06-06
Maintainer: Victor Kang [email protected]