Skip to main content

​
Overview

State management defines how plugins maintain context and data across their lifecycle. This configuration enables optimized resource usage and data sharing patterns.

​
State Management Modes

​
Global

The plugin starts at system initialization or on a specific trigger and maintains state across multiple tasks.
  • Available only to host plugins
  • Maintains context in shared memory on the host
  • Persists across multiple task executions
  • Single instance can serve multiple tasks
  • Machine learning models that remain loaded in memory
  • Caching layers for repeated operations
  • Shared databases or indices
  • Long-running analysis services

​
Isolated

The plugin is instantiated per task with no persistent state. Each task receives a fresh instance.
  • Complete isolation between tasks
  • No metadata or state persists after task completion
  • Simplest lifecycle model
  • Guaranteed clean state for each execution
  • Stateless analysis operations
  • Single-task workflows
  • Quick analysis tasks

​
Context Aware

The plugin can share context and data with specific plugins or tasks through controlled mechanisms.
  • Selective state sharing
  • Context propagation between related tasks
  • Controlled data flow between plugins
  • Balance between isolation and collaboration
  • Multi-stage analysis pipelines
  • Plugins that build on previous analysis results
  • Coordinated analysis workflows
  • Data aggregation across related tasks