Docker Health Skill

Verifies all 7 Docker services are healthy with prominent embedding model information display.

Usage

bash

/docker-health

Instructions

When to Use This Skill

MUST use when:

•User asks about Docker service health or infrastructure status
•Before starting development (verify infrastructure is ready)
•Embeddings aren't working (check TEI connectivity)
•After docker compose up (verify all services started correctly)
•Troubleshooting service failures or connectivity issues

DO NOT use when:

•Checking individual application logs (use docker logs directly)
•Deploying new containers (use docker compose up -d)
•Managing Docker volumes or networks (use Docker CLI directly)

Required Execution Method

MUST run the health check script FIRST:

bash

bash .claude/skills/docker-health/scripts/health-check.sh

This script automatically:

•Displays embedding model information banner (ALWAYS shown first)
•Checks all 7 services with color-coded output
•Shows metadata (queue stats, vector counts, response times)
•Provides suggested remediation commands for failures
•Exits 0 if all healthy, exits 1 if any unhealthy

When to Spawn docker-debugger Agent

MUST spawn the agent when:

•Script exits 1 (services are unhealthy)
•User requests detailed diagnostics or root cause analysis
•Multiple services are failing (need comprehensive log analysis)
•Network connectivity issues suspected (remote TEI unreachable)

Example trigger:

code

Script output: "Status: Degraded - fix unhealthy services ✗"
→ Spawn docker-debugger agent for deep diagnostics

Manual Service Checks (Alternative)

If script is unavailable or user needs individual service checks, use the detailed commands below.

Display Embedding Model Information First

Display embedding model information at the top of output to provide immediate context about the TEI service configuration. Use this banner format:

code

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
   EMBEDDING MODEL INFORMATION
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
✓ Model: Qwen/Qwen3-Embedding-0.6B
✓ Service: Hugging Face Text Embeddings Inference (TEI)
✓ Location: steamy-wsl (100.74.16.82:52000)
✓ Hardware: RTX 4070 GPU
✓ Vector Dimension: 1024
✓ Endpoints: /embed (native), /v1 (OpenAI-compatible)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Services to Check (7 total)

Check these services in order:

1. Firecrawl API (Port 53002)

bash

STATUS=$(curl -s --max-time 5 http://localhost:53002/health || echo "ERROR")
if [ "$STATUS" != "ERROR" ]; then
  echo "✓ firecrawl (53002) - healthy"
else
  echo "✗ firecrawl (53002) - unhealthy [Connection refused]"
  echo "  → docker restart firecrawl"
  echo "  → docker logs firecrawl --tail 50"
fi

2. Embedder Daemon (Port 53000)

bash

# Health check
HEALTH=$(curl -s --max-time 5 http://localhost:53000/health || echo "ERROR")

# Queue status
STATUS=$(curl -s --max-time 5 http://localhost:53000/status || echo "ERROR")

if [ "$HEALTH" != "ERROR" ]; then
  # Extract queue stats from JSON response
  PENDING=$(echo "$STATUS" | jq -r '.pending // 0')
  PROCESSING=$(echo "$STATUS" | jq -r '.processing // 0')
  echo "✓ embedder (53000) - healthy [${PENDING} pending, ${PROCESSING} processing]"
else
  echo "✗ embedder (53000) - unhealthy [Connection refused]"
  echo "  → docker restart firecrawl-embedder"
  echo "  → docker logs firecrawl-embedder --tail 50"
fi

3. Patchright (Port 53006 - Internal)

bash

STATUS=$(docker inspect firecrawl-playwright --format '{{.State.Status}}' 2>/dev/null || echo "ERROR")
if [ "$STATUS" = "running" ]; then
  echo "✓ patchright (53006) - running"
else
  echo "✗ patchright (53006) - $STATUS"
  echo "  → docker restart firecrawl-playwright"
  echo "  → docker logs firecrawl-playwright --tail 50"
fi

4. Qdrant (Port 53333)

bash

# Health check
HEALTH=$(curl -s --max-time 5 http://localhost:53333/health || echo "ERROR")

# Collection stats
COLLECTION=$(curl -s --max-time 5 http://localhost:53333/collections/firecrawl 2>/dev/null || echo "ERROR")

if [ "$HEALTH" != "ERROR" ]; then
  if [ "$COLLECTION" != "ERROR" ]; then
    VECTORS=$(echo "$COLLECTION" | jq -r '.result.vectors_count // 0')
    echo "✓ qdrant (53333) - healthy [${VECTORS} vectors]"
  else
    echo "✓ qdrant (53333) - healthy [no collections yet]"
  fi
else
  echo "✗ qdrant (53333) - unhealthy [Connection refused]"
  echo "  → docker restart firecrawl-qdrant"
  echo "  → docker logs firecrawl-qdrant --tail 50"
fi

5. Redis (Port 53379 - Internal)

bash

PING=$(docker exec firecrawl-redis redis-cli -p 53379 PING 2>/dev/null || echo "ERROR")
if [ "$PING" = "PONG" ]; then
  echo "✓ redis (53379) - healthy"
else
  echo "✗ redis (53379) - unhealthy"
  echo "  → docker restart firecrawl-redis"
  echo "  → docker logs firecrawl-redis --tail 50"
fi

6. RabbitMQ (Internal)

bash

HEALTH=$(docker inspect firecrawl-rabbitmq --format '{{.State.Health.Status}}' 2>/dev/null || echo "ERROR")
if [ "$HEALTH" = "healthy" ]; then
  echo "✓ rabbitmq - healthy"
elif [ "$HEALTH" = "starting" ]; then
  echo "⚠ rabbitmq - starting (wait 30-60s)"
else
  echo "✗ rabbitmq - $HEALTH"
  echo "  → docker restart firecrawl-rabbitmq"
  echo "  → docker logs firecrawl-rabbitmq --tail 50"
fi

7. Remote TEI (100.74.16.82:52000)

bash

INFO=$(curl -s --max-time 5 http://100.74.16.82:52000/info || echo "ERROR")
if [ "$INFO" != "ERROR" ]; then
  MODEL=$(echo "$INFO" | jq -r '.model_id // "unknown"')
  echo "✓ remote-tei (100.74.16.82:52000) - healthy [${MODEL}]"
else
  echo "✗ remote-tei (100.74.16.82:52000) - unreachable"
  echo "  → Check TEI service on steamy-wsl"
  echo "  → ping -c 3 100.74.16.82"
  echo "  → Verify firewall/Tailscale ACLs"
fi

Summary

Display final summary:

bash

echo ""
echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
HEALTHY_COUNT=$(grep -c "✓" <<< "$OUTPUT")
TOTAL_COUNT=7
echo "Services: ${HEALTHY_COUNT}/${TOTAL_COUNT} healthy"

if [ $HEALTHY_COUNT -eq $TOTAL_COUNT ]; then
  echo "Status: All systems operational ✓"
else
  echo "Status: Degraded - fix unhealthy services ✗"
fi
echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"

Color Coding

•Green ✓: Healthy, responding < 1s
•Yellow ⚠: Degraded, responding 1-5s or high queue
•Red ✗: Unhealthy, not responding or error

Metadata Display

•Embedder: Queue stats [X pending, Y processing]
•Qdrant: Vector count [N vectors]
•TEI: Model name [Qwen3-Embedding-0.6B]
•All services: Response times for HTTP checks

Integration

This skill can spawn the docker-debugger agent for deep diagnostics:

•If services are unhealthy, suggest using the agent for root cause analysis
•Agent provides log analysis, network diagnostics, and remediation steps
•Skill provides quick health overview, agent provides detailed investigation

Output Format

Uses project's theme.ts utilities for consistent styling:

•Green ✓ for healthy services
•Red ✗ for unhealthy services
•Yellow ⚠ for degraded/starting services
•Dim text for suggested commands