Coder integrates Claude Code into Agent Relay to run agents on customer infrastructure
The integration lets regulated enterprises run Claude Code agent execution inside their own controlled environments (audit logs, egress policies, sandboxing) while Anthropic retains account management and billing.
In this brief: 3 sections 1 min read
Workspaces are sandboxed, ephemeral, and scoped to a single Claude Code session.
Tool calls and filesystem/credentials remain inside customer-run workspaces.
Anthropic handles account administration, billing and the agent loop; execution stays on customer infrastructure.
Egress policies and audit trails are inherited from the workspace model.
Each run produces correlated audit records of workspace lifecycle and Claude session.
Aimed at regulated enterprises requiring clear records of where code runs and what it can access.
Integration leverages Claude Code’s publicly documented self-hosted environments.
No additional deployment required for existing Agent Relay customers; Coder plans further extensions.