Skip to main content
The Kubiya Control Plane API is a multi-tenant AI agent orchestration and management platform built with FastAPI, Temporal, and PostgreSQL.

Features

  • Multi-tenant Architecture: Manage multiple projects, teams, and agents
  • Workflow Orchestration: Temporal-based workflow execution
  • Flexible Agent Runtime: Support for multiple agent types and skills
  • Policy Enforcement: OPA-based policy engine for agent governance
  • Scalable Workers: Distributed worker architecture for agent execution
  • Context Management: Environment and team-specific context handling
  • LLM Integration: Support for multiple LLM providers via LiteLLM
  • Comprehensive APIs: RESTful APIs for all platform features

API Resources

The Control Plane API provides the following resources:
ResourceEndpointDescription
Agents/api/v1/agentsAgent management and configuration
Analytics/api/v1/analyticsUsage analytics and metrics
Context/api/v1/contextContext management for agents
Context Graph/api/v1/context-graphKnowledge graph and relationships
Environments/api/v1/environmentsEnvironment configuration
Execution Environment/api/v1/execution-environmentExecution environment resolution
Executions/api/v1/executionsExecution tracking and history
Health/api/healthHealth and readiness checks
Integrations/api/v1/integrationsThird-party integrations
Jobs/api/v1/jobsJob management and scheduling
Models/api/v1/modelsLLM model configuration
Policies/api/v1/policiesAccess policies and permissions
Presence/api/v1/presenceUser presence tracking
Projects/api/v1/projectsMulti-project management
Runners/api/v1/runnersRunner management
Runtimes/api/v1/runtimesAgent runtime types
Secrets/api/v1/secretsSecret management
Skills/api/v1/skillsTool sets and capabilities
Task Queues/api/v1/task-queuesTask queue management
Tasks/api/v1/tasksTask planning and execution
Teams/api/v1/teamsTeam configuration
Templates/api/v1/templatesTemplate compilation
Worker Queues/api/v1/worker-queuesWorker queue management
Workers/api/v1/workersWorker management
Workflows/api/v1/workflowsWorkflow orchestration

Architecture

Workflow Orchestration

The platform uses Temporal for reliable workflow execution:
  • Durable execution with automatic retries
  • Activity-based task decomposition
  • Support for long-running workflows
  • Built-in observability and monitoring

Multi-tenancy

  • Projects: Top-level isolation boundary
  • Environments: Isolated execution contexts within projects
  • Teams: Collaborative agent groups
  • Agents: Individual agent instances

Worker Architecture

Workers pull tasks from environment-specific queues and execute agent workflows using Temporal.