Agent workflow / Overview
Agent workflow
Owned workflow index
Your agent workflows
Open any agent workflow you own, or begin with a clean workspace.
No agent workflows yet
Create your first executable agent workflow. Nothing has been generated or opened for you.
Your agent workflow, grounded in evidence.
Design executable nodes, connect research, and validate decisions before your next run.
Workflow nodes
Executable units and decisions in your agent workflow
Research suggestions
Generated from your research library
Model comparison and workflow runs
Choose exactly one execution model for each custom instruction. Every output or provider error is saved to this workflow.
Executable graph
Workflow graph
Make every node contract, decision, and handoff explicit, testable, and traceable to a citation.
Node map and handoff chain
Execute agent workflow
Complete each required contract input, run one provider-backed node, or execute the ordered workflow.
Instruction design
Prompt components
Attach evidence to each prompt component so reusable instructions stay defensible.
Workflow instructions
● Backend persistedEvaluation
Sandbox scenarios
Inspect and refine persisted sample checks, then run selected edge cases without leaving the sandbox.
Backend-loaded evidence base
Research library
Curated reference data loaded from the Atlas backend; sources are not presented as user-owned research.
Change-aware review
System review
Review what changed since the last sign-off, then record a new decision.
Persisted review history
Backend-backed builder
Workflow agent
Chat with the configured model, or invoke explicit tools whose results are persisted against this agent workflow.
Agent conversation
Messages and provider errors persist on reload.