Neo4j-based organizational knowledge graph storing resources, relationships, and execution history.
Quick Start
# View graph statistics
kubiya graph stats
# List all node labels
kubiya graph labels
# Execute custom Cypher query
kubiya graph query --query "MATCH (n:Agent) RETURN n LIMIT 10"
📊 Graph Statistics
Total Nodes: 1,247
Total Relationships: 3,891
Node Types:
- Agent: 42
- Team: 15
- Environment: 8
- Skill: 127
- Execution: 892
- User: 163
# List available integrations
kubiya graph integrations list
📊 Available Integrations (32)
INTEGRATION DESCRIPTION
aws Amazon Web Services cloud platform
github GitHub repository management
slack Slack team communication
kubernetes Kubernetes cluster management
terraform Infrastructure as code
datadog Monitoring and analytics
pagerduty Incident management
jira Issue tracking and project management
[... 24 more integrations]
Graph Statistics
Get Overall Statistics
# View overall statistics
kubiya graph stats
# Stats for specific integration
kubiya graph stats --integration github
Nodes
List Nodes
# List all nodes
kubiya graph nodes list
# Filter by label
kubiya graph nodes list --label Agent
kubiya graph nodes list --label Team
# Limit results
kubiya graph nodes list --limit 100
Get Node Details
# View specific node
kubiya graph nodes get <node-id>
Get Node Relationships
# View node relationships
kubiya graph nodes relationships <node-id>
Search Nodes
# Search by property
kubiya graph nodes search \
--label Agent \
--property-name name \
--property-value "deploy"
Integrations
List all integrated data sources that contribute to the knowledge graph:
# List all integrated data sources
kubiya graph integrations list
Custom Queries
Run custom Cypher queries to analyze your organizational knowledge graph:
# Run Cypher query
kubiya graph query --query "MATCH (n:Agent) RETURN n LIMIT 10"
# Find agents with specific skill
kubiya graph query --query "MATCH (a:Agent)-[:HAS_SKILL]->(s:Skill) WHERE s.name = 'kubernetes' RETURN a.name"
# Analyze team composition
kubiya graph query --query "MATCH (t:Team)-[:HAS_MEMBER]->(a:Agent) RETURN t.name, count(a) as member_count"
Schema Discovery
List Labels
# List all node labels
kubiya graph labels
📋 Node Labels (12)
- Agent
- Team
- Environment
- Skill
- Policy
- Execution
- User
- Integration
- Secret
- WorkerQueue
- Job
- Project
List Relationship Types
# List all relationship types
kubiya graph relationship-types
🔗 Relationship Types (18)
- HAS_SKILL
- HAS_MEMBER
- BELONGS_TO
- EXECUTED_BY
- USES_POLICY
- CREATED_BY
- HAS_INTEGRATION
- DEPLOYED_TO
- DEPENDS_ON
...
Subgraph Exploration
Extract a subgraph around a specific node:
# Get subgraph around node
kubiya graph subgraph --node-id <node-id> --depth 2
Intelligent Search
AI-powered natural language search for the context graph using Claude:
# Simple search
kubiya graph search "Find all production environments in AWS"
# Search with streaming
kubiya graph search "Show me critical security issues" --stream
# Advanced search with parameters
kubiya graph search "Find Kubernetes clusters" \
--max-turns 10 \
--model kubiya/claude-opus-4 \
--temperature 0.5 \
--integration AWS
# Continue a conversation
kubiya graph search "Tell me more about the first one" --session abc-123-def
# JSON output
kubiya graph search "List all teams" --output json
🤖 AI-Powered Graph Search
🔍 Query: Find all production environments in AWS
Searching context graph...
✓ Found 3 production environments:
1. prod-us-east-1 (Environment)
Region: us-east-1
Status: active
Resources: 42 nodes
2. prod-us-west-2 (Environment)
Region: us-west-2
Status: active
Resources: 38 nodes
3. prod-eu-west-1 (Environment)
Region: eu-west-1
Status: active
Resources: 31 nodes
- AI-powered: Claude-based agent with specialized graph tools
- Flexible: Configurable model, temperature, and search parameters
- Smart tools: 10 graph operations (property search, relationships, subgraphs, etc.)
- Streaming: Real-time progress updates (default: enabled)
- Session continuity: Continue conversations with
--session flag
Optional Flags:
--stream - Enable real-time streaming (default: true)--model - LiteLLM model name (default: kubiya/claude-sonnet-4)--temperature - Model temperature 0.0-2.0 (default: 0.7)--max-turns - Maximum conversation turns 1-20 (default: 5)--integration - Filter by integration (AWS, Azure, Slack, etc.)--label - Filter by node label--semantic - Enable semantic search--cypher - Enable custom Cypher queries--output - Output format (text, json)
Use intelligent search for exploratory queries. For structured data retrieval, use kubiya graph nodes commands or custom Cypher queries.
Best Practices
- Use graph queries for resource discovery and pattern analysis
- Monitor agent activity and execution success rates
- Optimize graph queries with indexes and limits
- Archive old execution data periodically
- Implement regular Neo4j backups
- Use intelligent search for exploration, structured queries for automation
- Filter queries by integration or label for performance
- Set appropriate depth limits for subgraph queries
Command Reference
# Statistics
kubiya graph stats [--integration <name>]
# Nodes
kubiya graph nodes list [--label <type>] [--limit <n>]
kubiya graph nodes get <id>
kubiya graph nodes relationships <id>
kubiya graph nodes search --label <type> --property-name <key> --property-value <val>
# Integrations
kubiya graph integrations list
# Custom Queries
kubiya graph query --query "<cypher>"
# Schema
kubiya graph labels
kubiya graph relationship-types
# Subgraph
kubiya graph subgraph --node-id <id> --depth <n>
# Intelligent Search
kubiya graph search "<query>" [--stream] [--model <name>] [--temperature <n>]
kubiya graph search "<query>" --integration <name> --label <type>
kubiya graph search "<query>" --session <id> --max-turns <n>
Next Steps
