AgentWorks documentation

Documentation

Start with the operator journey, then use the subsystem references when you need implementation detail.

Start Here

Getting Started: install AgentWorks, complete first-launch setup, and create a first automation.
Build Your First Workflow: define an outcome, choose a worker, run it, review evidence, and improve the next run.

Product Areas

Workflow: workflow authoring, execution, scheduling, monitoring, Pulse, and Auto Improve.
Organization and Agents: delegation, Org Pulse, shared memory, and agent-to-agent coordination.
Core: providers, MCP, browser sessions, connectors, secrets, security, and shared runtime services.

docs/bugs/ is an incident archive. docs/refactor/ records implementation migrations. Neither folder is the recommended entry point for operators.

Placement Rules

Put a doc in workflow/ when it is primarily about workflow authoring, workflow execution, step configuration, or workflow-only UX.
Put a doc in multiagent/ when it is primarily about manager/worker delegation, multi-agent chat, or agent-to-agent coordination.
Put a doc in core/ when it applies across chat, workflow, and multi-agent modes or describes a foundational subsystem or integration.

Current Sections

Workflow

React Flow canvas and workflow UX
Running workflows and workflow manifests
Step config, execution modes, tool filtering, and runtime overrides
Validation, learning, evaluation, human feedback, and monitoring
Specialized workflow step types such as todo_task

Organization and Agents

Sub-agent delegation
Multi-agent chat architecture
Agent memory
Slash-command entry points for delegation

Core

Eventing, streaming, session propagation, auth, secrets, and workspace isolation
LLM provider configuration and resilience
Browser, MCP bridge, bot connectors, and external integrations
Platform plans and operational system docs

Validate Links

Run the documentation link check before merging documentation changes:

node scripts/check-doc-links.js