Cursor Unleashes Public Beta of TypeScript SDK: Embed Powerful AI Agents into Your Workflow

⚙️ Flexible Execution Modes

• local execution: Run Agents directly on your local machine for immediate feedback and debugging.

• Cursor Cloud: Agents operate within dedicated sandboxed virtual machines. This ensures that long-running tasks continue seamlessly even if your laptop lid is closed or your internet connection drops. Upon completion, agents can automatically open Pull Requests, push branches, or attach screenshots of their work.

• Self-Hosted Worker Nodes: For enterprises requiring full control, agents can be deployed on self-hosted infrastructure.

🛠️ Full-Spectrum Agent Capabilities

• Deep Context Understanding: Utilizing codebase indexing and semantic search to underStand complex project structures.

• Tool Integration: Connecting to external tools via MCP (Model Context Protocol) servers.

• Skill Management: Automatically loading skills from a designated directory and using Hooks to control agent loops.

• Task Delegation: The ability to spawn sub-agents to handle specific sub-tasks within a larger workflow.

💡 Key Use Cases

• CI/CD Integration: Trigger agents directly from continuous integration pipelines to automatically fix bugs, summarize changes, or handle routine mAIntenance.

• Product embedding: Embed the agent experience directly into customer-facing products, allowing end-users to interACT with AI for code generation or troubleshooting.

• Internal Tools: Several companies are already utilizing the SDK to build internal "no-code" query tools for their engineering teams.

💰 Availability