Cloud agent topics

Configure cloud agents, inspect runs, and troubleshoot setup using the topic links below. For your first standalone run, start with the Cloud agents quickstart.

Architecture and execution

• Execution architecture

• Cloud agents overview

• Inspecting and steering runs

• Codebase Context for cloud runs

• Environment tools and dependencies

• Container users and permissions

• Self-hosting and required network access

Models

• Model choice

• Selecting a model for a run

• Harness authentication

• Private inference options

• Current orchestration capabilities

Security and billing

• Security and encryption

• Cloud access and usage billing

• Standalone cloud secrets

Workflows

• Branching and coordination

• Orchestration patterns

• Connecting external tools with MCP

• Retrieving conversations and artifacts

• Environment access and credential boundaries

• PR review in GitHub Actions

• Test-generation instructions for recurring runs

• Splitting large refactors across agents

• Building an issue-triage workflow

• Scheduled dependency maintenance

• Scheduled documentation checks

Self-hosting

• Source code, transcripts, and storage

• Running agents in CI or existing hosts

• VPN and private-service access

• Source-control integrations

• Inference routing with self-hosting

• Pre-provisioned monorepos and managed storage

• Kubernetes security boundaries

Image attachments

• Interactive image attachments and CLI files

Related pages

• Cloud agents overview - Choose a standalone workflow
• Cloud agent session sharing - Inspect and steer a run