Cursor releases Agent SDK for code-defined agents
Original title: The Agent SDK
Teams can define agents in TypeScript and Markdown, run evals, and deploy them to Cursor or their own Node process.
Cursor released the Agent SDK, the framework it uses for managed agents such as Bugbot, Security Reviewer, and Rollouts. Developers define agents as code, with Markdown prompts and TypeScript tools, and can deploy them with agent-sdk deploy or run each agent as a single Node process. The SDK supports event-driven workflows, subagents, evals, and event replay. Cursor says it is generally available and cites PGA Tour as a user.
- An internal kb-reviewer agent was built in about a day and runs on every main-repo pull request.
- Agents can respond to pull request events from GitHub, GitLab, or Bitbucket.
- The sample agent configuration names the model composer-2.5.
- Secrets and integrations are managed from the Cursor dashboard.
- Vendors
- Cursor
- Products
- Agent SDK
- Organizations
- PGA Tour
- AI models
- composer-2.5
From the original article
3 min read at cursor.com
Software teams use agents defined as code to handle recurring work across the development lifecycle. These agents connect to the places where a team works and act when relevant events occur, whether that means reviewing a pull request, executing a compliance workflow, or investigating an incident. The most useful agents are tailored to how a particular team operates.
By torutofu · cursor.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.