Cline is now available in the AI SDK harness layer
Vercel’s AI SDK harness layer now supports Cline, enabling developers to integrate and switch between multiple coding-agent runtimes via a unified interface without modifying application code.
The AI SDK harness layer provides a standardized way to execute established coding-agent runtimes through a single interface, allowing developers to swap between tools like Cline, Claude Code, or Pi without altering their underlying codebase. Vercel has introduced a new adapter, @ai-sdk/harness-cline, developed in partnership with the Cline team, to integrate Cline into this system. This adapter enables Cline to operate within the host process while using the sandbox exclusively for filesystem and shell interactions, mirroring the approach used by the existing Pi harness adapter. The update expands the supported list of harnesses to include Cline alongside existing options such as Claude Code, Codex, and Deep Agents, with additional tools expected to be added in future releases.
Cline’s integration follows Vercel’s design for the harness layer, which prioritizes minimal overhead by avoiding the installation of bridge processes inside sandboxes. Instead, Cline relies on the sandbox as a remote filesystem and shell, ensuring that only its tools execute within the isolated environment. This approach aligns with the existing Pi harness, which also operates under the same constraints. Developers can now leverage Cline’s capabilities within the AI SDK harness layer, maintaining consistency across runtime environments without requiring code changes. The documentation for the Cline harness is available to guide implementation.
The addition of Cline brings the total number of supported harnesses in the AI SDK to seven, including Claude Code, Codex, Deep Agents, Grok Build, OpenCode, and Pi. Vercel has indicated that more harnesses will be added over time, reflecting the growing demand for flexible, multi-runtime workflows in AI-driven development. The unified interface simplifies the process of testing and deploying different coding agents, as developers can switch between them with minimal configuration. This update underscores Vercel’s commitment to expanding the AI SDK’s capabilities while maintaining a consistent developer experience.
To begin using Cline within the AI SDK harness layer, developers must install the @ai-sdk/harness-cline package and configure their environment to recognize the new runtime. The full list of supported harnesses is documented on Vercel’s official resources, providing clear guidance on integration and usage. This update does not introduce new pricing or licensing requirements, as the harness layer remains part of the existing AI SDK framework. Developers are encouraged to review the Cline harness documentation for technical details and best practices.