conductor-oss/conductor

Features

• Durable Workflow Engines - Persists process history to ensure reliable task completion across infrastructure failures and system restarts.
• Durable Workflow Execution Engines - Persists execution state at every step to ensure processes survive failures while automatically handling retries and compensation.
• Workflow Engines - Provides a core engine for executing background tasks across multiple programming languages.
• Workflow Orchestration - Manages the full lifecycle of tasks including scheduling, execution, and error handling within durable workflows.
• Agent State Persistence - Saves the entire execution state of an agent to ensure reliability across system restarts and failures.
• AI Agent Orchestration Frameworks - Provides a specialized runtime for managing autonomous agent loops, tool discovery, and model-based reasoning.
• Durable AI Agent Orchestration - Maps agent concerns like planning and memory management to durable workflow primitives for reliable execution.
• Durable Execution Persistence - Saves the state of every workflow and task to storage, ensuring progress is recoverable after system crashes or infrastructure failures.
• Workflow Compensation Systems - Implements automated rollback and cleanup logic to handle failure scenarios within durable workflow executions.
• Workflow Orchestration Engines - Durable workflow engines run workflows generated at runtime by passing definitions directly to the engine, ensuring every generated step remains durable, observable, and retryable.
• Workflow Persistence Engines - Durable workflow engines save workflow and task execution state to a database at every step to ensure progress is never lost during infrastructure failures or server restarts.
• AI Agent Orchestrators - Connects language model providers with external tools and human approval steps to build automated intelligent pipelines.
• AI Workflow Orchestration - Integrates model inference and tool calling into workflows to connect AI capabilities with business logic.
• Autonomous Agent Loops - Combines reasoning tasks with conditional execution logic to repeatedly plan, act, and observe until completion.
• Distributed Task Orchestrators - Manages task lifecycles, worker polling, and parallel execution branches across heterogeneous computing environments.
• State Machine Orchestrators - Executes complex business logic by transitioning through defined states while persisting the entire execution history.
• Workflow Blueprints - Durable workflow engines specify task sequences, input parameters, and runtime policies like timeouts and compensation flows to create durable business logic blueprints.
• Workflow Versioning - Durable workflow engines maintain execution consistency by using immutable snapshots of definitions, ensuring active workflows complete with original logic while new executions adopt updated versions.
• Agent Failure Mitigation - Executes automated compensation workflows and retry policies to resolve errors during complex agentic operations.
• Human Approval Gates - Requires human approval before executing actions to prevent unauthorized real-world consequences.
• Multi-Agent System Composition - Nests sub-agents within parallel structures to delegate specialized tasks and synthesize results.
• Distributed Task Schedulers - Automates recurring business operations with precise timing, retry policies, and backoff strategies.
• Event-Driven Automation Platforms - Triggers and resumes complex business processes based on external signals and message bus events.
• Workflow Schedulers - Durable workflow engines define workflow execution schedules using expressions with second-level precision to trigger automated tasks at specific intervals or times.
• Distributed Transaction Managers - Models complex operations as sagas that automatically execute undo logic in reverse order when a step fails.
• Event-Driven Architectures - Durable workflow engines register event handlers that automatically trigger the execution of specific workflows when messages arrive on configured queues or topics from external systems.
• Missed Execution Handlers - Durable workflow engines trigger all missed workflow executions automatically upon system restart to ensure data consistency for critical processes like billing or data processing.
• Retry Strategies - Durable workflow engines manage transient failures by defining task retry behavior using fixed, linear, or exponential backoff strategies to reduce load on downstream services.
• Workflow Monitoring - Durable workflow engines track the lifecycle of workflow instances as they transition through states like running, completed, failed, or paused during their execution.
• Workflow Task Definitions - Durable workflow engines create or update task definitions through the visual interface, command-line tools, or APIs to maintain consistent workflow logic.
• Agentic Tool Orchestration - Manages tool discovery, planning, and task execution through standardized interfaces for agentic operations.
• Autonomous Agent Creation - Builds autonomous agent loops using conditional logic to repeatedly think, act, and evaluate progress.
• Human Approval - Inserts human-in-the-loop tasks into workflows to pause execution until a user approves planned actions.
• Workflow-as-a-Tool Exposure - Exposes existing workflows as tools through a gateway, allowing external agents to discover and invoke them.
• Data Engineering - Orchestration engine for complex business workflows.
• General Purpose Orchestration - Microservices orchestration platform for building scalable, durable processes.
• Distributed Task Queues - Workers retrieve pending tasks from a centralized queue system, enabling horizontal scaling and decoupling of execution logic.
• System Task Executors - Durable workflow engines run built-in system tasks directly on the server to handle common operations like HTTP requests, event publishing, and inference without writing custom worker code.
• Worker Scaling Strategies - Durable workflow engines increase worker capacity horizontally by running multiple instances, applying rate limits, and using domain isolation to route tasks to specific worker groups.
• Concurrent Execution Managers - Durable workflow engines design workflows to handle concurrent executions by ensuring idempotency or adjusting intervals when workflow duration exceeds the frequency of the scheduled trigger.
• Dynamic Task Graphs - Constructs and modifies execution paths at runtime based on previous outputs, enabling flexible branching and parallel processing.
• Failure Handling Policies - Durable workflow engines manage task-level failures by marking steps as optional, signaling terminal errors to bypass retries, or setting specific timeouts to prevent blocking workflow execution.
• Microservices Process Coordinators - Chains multiple service calls and parallel operations into a single, observable, and versioned execution pipeline.
• Reliable Task Queues - Durable workflow engines ensure at-least-once task delivery by automatically redelivering tasks if a worker fails to report completion within the configured response timeout period.
• Automated Output Evaluation - Automates quality and safety checks by using a secondary model to judge outputs.
• External Tool Execution - Runs tools on external servers with durable execution and automatic retries to ensure reliability.
• Human Interaction Management - Handles long-running human interaction tasks by configuring explicit timeouts and escalation patterns.
• Model Task Retries - Retries failed model tasks automatically using configurable backoff strategies while preserving workflow state.
• Token Optimization Strategies - Persists model call results to storage to prevent redundant execution and minimize token usage during retries.
• Event-Driven Workflow Triggers - Resumes or initiates workflow execution by listening for external signals and messages from integrated message brokers.
• Task Worker Configurations - Durable workflow engines set task execution parameters including retry policies, response timeouts, and concurrency limits to manage how workers process tasks within a workflow.
• Retry Policies - Durable workflow engines set task-level durability settings including retry policies, backoff strategies, and timeout thresholds to handle transient failures and long-running operations.
• Task Templates - Durable workflow engines create reusable task configurations including retry policies, timeout limits, and input parameters to standardize behavior across multiple workflow definitions.
• Timeout Policies - Durable workflow engines determine whether a task should be retried, trigger a workflow failure, or log an alert when execution limits are exceeded.
• System Monitoring - Durable workflow engines track queue sizes and worker activity using the interface or API to detect backlogs and capacity issues in real time.
• Event-Driven Integration Platforms - Connects disparate microservices by triggering workflows from events and coordinating asynchronous task execution.
• AI Agent Workflow Definition - Configures agent workflows using data definitions that orchestrate tool discovery and planning.
• Conditional Result Review - Triggers human review of tool execution results only when specific conditions are met.
• External Tool Call Timeouts - Configures timeouts for external tool calls to trigger retries while maintaining execution idempotency.
• External Tool Discovery - Queries external servers at runtime for tool names and parameter schemas to enable dynamic capabilities.
• Multimodal Generation Pipelines - Integrates native system tasks to generate and process images, audio, video, and documents within automated execution pipelines.
• Workflow Scheduling Constraints - Durable workflow engines restrict the active period of a workflow schedule by defining specific start and end timestamps to automate the lifecycle of time-bound tasks.
• Event Brokers - Durable workflow engines send messages to external message brokers by configuring an event task with a specific destination and input parameters.
• Worker Pool Management - Durable workflow engines adjust worker pool capacity by tuning polling intervals, thread counts, and concurrency limits to route tasks to specific worker environments efficiently.
• Idempotency Mechanisms - Durable workflow engines reject duplicate task completion callbacks automatically to prevent redundant state transitions or multiple workflow advancements for the same task.
• Idempotency Patterns - Ensures reliable operations by requiring workers to handle duplicate task identifiers, preventing redundant side effects.
• Execution History Auditors - Durable workflow engines search and retrieve the execution history of scheduled workflows to monitor their status, including timestamps and success or failure states.

Star history