[Release] openai/openai-agents-python v0.14.0: v0.14.0

OpenAI's openai-agents-python v0.14.0 introduces Sandbox Agents, a beta feature enabling agents to run in persistent, isolated workspaces with file access, command execution, and cross-run continuity. The release includes multiple execution backends (local, Docker, hosted providers), workspace management with snapshots and resume capabilities, sandbox memory for learning across runs, and comprehensive examples. Runtime enhancements provide tracing, session lifecycle management, and seamless integration with the existing Agent and Runner framework.

Key Points

• •SandboxAgent extends Agent with sandbox defaults, manifests, and capabilities for persistent workspace operations
• •Multiple execution backends supported: UnixLocalSandboxClient, DockerSandboxClient, and hosted providers (Blaxel, Cloudflare, Daytona, E2B, Modal, Runloop, Vercel)
• •Workspace manifests define fresh-workspace contracts including files, directories, Git repos, environment, users, groups, and mounts
• •Sandbox memory capability enables agents to learn from prior runs with read-only/generate-only modes and progressive disclosure
• •Workspace snapshots and serialized session state allow reconnecting to existing work or seeding fresh sandboxes from saved contents
• •Remote storage mount support for S3, Cloudflare R2, Google Cloud Storage, Azure Blob Storage, and S3 Files with provider-specific strategies
• •SandboxRunConfig provides per-run sandbox wiring for client creation, session injection, manifest overrides, and materialization concurrency limits
• •Built-in capabilities include shell access, filesystem editing, image inspection, skills, memory, and compaction
• •Runner-managed sandbox lifecycle handles preparation, capability binding, session management, state serialization, and resume behavior
• •Comprehensive examples suite covering local/Docker runners, coding tasks, multi-agent workflows, and domain-specific tutorials (tax-prep, healthcare, dataroom QA)

Workflow Diagram

Concepts