Last updated: February 21, 2026

Data Handling

This policy explains what data flows through Swarm during issue orchestration and how teams can control that flow.

Data flow scope

Swarm processes repository and issue context to plan and execute requested engineering tasks.

Organizations decide which repositories and integrations are connected and which users can run automations.

Inputs can include issue metadata, repository code context, run configuration, and reviewer feedback.

Outputs can include generated code changes, test artifacts, run events, and pull request summaries.

Depending on model configuration, execution data may be processed by third-party AI providers selected by your organization.

Customers are responsible for reviewing provider terms and choosing model routes appropriate for their compliance requirements.

Retention windows for logs and artifacts are tied to workspace settings and operational needs.

Where supported, organizations can configure shorter retention windows for sensitive workflows.

Swarm supports event-level logging that helps teams inspect who initiated runs and what actions were taken.

These records can be used for engineering governance, review traceability, and incident investigations.

For production rollouts, combine Swarm controls with your own branch protections, CI policies, and approval workflows.

If your organization needs additional contractual terms, align with your account owner before deployment.

Questions about these policies should be routed through your Swarm workspace support channel.