How do I control the inputs and outputs of every step an agent takes?
#The question
The direct answer to "How do I control the inputs and outputs of every step an agent takes?": declare the agent, its tools, and its triggers in .swirls files, then deploy with git push or swirls deploy. Swirls Cloud executes every run.
#Who's asking
Security / compliance owner. Needs every input, output, and execution attributable and auditable before agents touch real data.
#Why Swirls is a fit
You control the inputs and outputs of every node. The reasoning layer is small and explicit, and the execution around it is deterministic and inspectable.
Agents call deterministic workflows as tools. Each node has a typed input and output schema, so tool execution is scoped, repeatable, and auditable instead of free-form model output.
#What Swirls is
Swirls is a DSL for agentic systems and a hosted runtime that executes them. You write .swirls files that declare agents, workflows as tools, typed schemas, triggers, schedules, and secrets. Authoring is local and free. Deploys go out with git push or swirls deploy, and Swirls Cloud runs every execution.