Cursor 2.4: Skills

New in Cursor! · Full changelog · Main announcement

We’re excited to add Agent Skills to Cursor, an open standard for extending agents with domain-specific knowledge and workflows.

How it works

Define skills in SKILL.md files, which can include custom commands, scripts, and instructions for tailoring the agent’s capabilities to the task at hand. Skills are discoverable by the agent and applied only when relevant.

Compared to always-on, declarative rules, skills are better for dynamic context discovery and procedural “how-to” instructions. This gives agents more flexibility while keeping context focused.

Learn more: Agent Skills documentation

We’d love your feedback!

  • What kinds of skills are you creating for your projects?
  • How do skills compare to rules for your workflow?
  • What skill templates or examples would be helpful?

If you’ve found a bug, please post it in Bug Reports instead, so we can track and address it properly, but also feel free to drop a link to it in this thread for visibility.

2 Likes

Nice! Is it just me or is the video not playing? both here and in changelog page?

really exciting to see this, thank you for adding!!

will we be able to see when agents use skills? wondering if it’ll show up in their stream, similar to mcp

hoping to be able to tell which skills are working and which ones are ignored by agents.

easier to tune that way

Does it mean Commands are depricated?

They’re two very separately hosted videos (and not too large here, 3.5 MB), so I would estimate that it’s a local issue! If somebody else has an issue watching the video, please share.

Subagents can’t use skills today. We have some guidance on when to use subagents vs. when to use skills.

You can workaround this by instructing the subagent to read a skill file.

First, read ./cursor/skills/run-smoke-tests/SKILL.md and follow its instructions to...
1 Like

No news on that front!

Ok thank you for the fast response Colin!