Managing Qdrant

Qdrant Management Skill

Monitor, analyze, and optimize Qdrant vector database instances safely.

MANDATORY: Discovery-First Pattern

Always check cluster health and list collections before any query or indexing operations. Never assume collection names, vector dimensions, or payload schemas.

Phase 1: Discovery

#!/bin/bash

QDRANT_URL="${QDRANT_URL:-http://localhost:6333}"
QDRANT_AUTH="${QDRANT_API_KEY:+api-key: $QDRANT_API_KEY}"

qd_get() {
    curl -s ${QDRANT_AUTH:+-H "$QDRANT_AUTH"} "$QDRANT_URL$1"
}

echo "=== Health Check ==="
qd_get "/healthz"

echo ""
echo "=== Telemetry / Version ==="
qd_get "/telemetry" | python3 -c "
import sys, json
data = json.load(sys.stdin)
app = data.get('result', {}).get('app', {})
print(f\"Version: {app.get('version','?')}\")
print(f\"Startup: {app.get('startup','?')}\")
collections = data.get('result', {}).get('collections', {})
print(f\"Total collections: {collections.get('number_of_collections','?')}\")
" 2>/dev/null

echo ""
echo "=== Collections ==="
qd_get "/collections" | python3 -c "
import sys, json
data = json.load(sys.stdin)
for c in data.get('result', {}).get('collections', []):
    print(f\"Collection: {c['name']}\")
" 2>/dev/null

echo ""
echo "=== Collection Details ==="
for coll in $(qd_get "/collections" | python3 -c "
import sys, json
for c in json.load(sys.stdin).get('result',{}).get('collections',[]):
    print(c['name'])
" 2>/dev/null); do
    qd_get "/collections/$coll" | python3 -c "
import sys, json
r = json.load(sys.stdin).get('result', {})
config = r.get('config', {})
params = config.get('params', {})
vectors = params.get('vectors', {})
if isinstance(vectors, dict) and 'size' in vectors:
    print(f\"  {r.get('name','?')}: dim={vectors['size']} distance={vectors.get('distance','?')} points={r.get('points_count',0)} status={r.get('status','?')}\")