Does Cursor support skills?

Yes. Cursor added Agent Skills in version 2.4, in both the editor and the command-line tool. A skill is a folder with a `SKILL.md` file holding a name, a description and the instructions. Cursor loads skills from your project and user folders, and the agent applies one when a request matches or when you call it with `/` and its name.

Where do Cursor skills go?

Project skills go in `.cursor/skills/` or `.agents/skills/` inside your repository, each in its own folder with a `SKILL.md` file. Skills you want in every project go in the same folders in your home directory. Cursor also loads skills from `.claude/skills/` and `.codex/skills/`, so skills written for Claude Code or Codex are picked up too.

How do I create a skill in Cursor?

Type `/create-skill` in Agent chat and describe one job, its inputs and the output you want. Cursor's built-in skill walks you through naming, structuring and saving it. You can also write the folder and `SKILL.md` file yourself. Either way, check the description, add what to do when information is missing, and test it with three cases.

What's the difference between Cursor rules and skills?

Rules are short guidelines Cursor includes in every conversation, or in conversations about matching files, such as a coding convention. Skills are longer procedures the agent uses when a job calls for them, such as deploying to staging. Use a rule when a short instruction is enough, and a skill when the agent needs a detailed, repeatable process.

Can I use my Claude skills in Cursor?

Often, yes. Cursor reads the same `SKILL.md` format and loads skills from Claude's `.claude/skills/` folders, so many carry across without changes. Extra settings, scripts and references to another tool's connectors may behave differently, so run your test cases in Cursor first. Contexta avoids keeping two copies by delivering one maintained skill to both tools through their connections.

How do I share Cursor skills with my team?

Commit project skills to your repository, and everyone who works in it gets them. On Teams and Enterprise plans, you can also publish a personal skill to your team's marketplace from Customize, then Skills. Teammates choose whether to install it, and you push updates with Sync changes. Shared skills stay inside Cursor, so colleagues using other AI tools need their own copy.

Cursor Skills: A Practical Guide

Quick Answer

Cursor skills are folders of instructions, built around a SKILL.md file, that teach Cursor's agent how to do one specific job. Cursor finds them in your project or your user folders. The agent picks one up when a request matches, or you choose it by typing / and its name. Cursor follows the open Agent Skills standard, so skills written for other tools often carry across.

What Are Cursor Skills?

Cursor added Agent Skills in version 2.4, released on 22 January 2026, in both the editor and the command-line tool (Cursor changelog: 2.4). Cursor describes a skill as a portable, version-controlled package that teaches the agent how to do a domain-specific task (Cursor docs: Agent Skills).

Think of a skill as a job card for a capable new starter: the steps, the format and the checks for one job, so the work comes out the same way whoever asks.

Each skill is a folder containing a SKILL.md file. The file opens with two required fields: a name, in lowercase letters, numbers and hyphens and matching the folder name, and a description saying what the skill does and when to use it. The agent reads the description to decide whether the skill is relevant. Optional scripts/, references/ and assets/ folders hold code the agent can run, longer reference material and templates. Cursor's docs call Agent Skills an open standard; What Are Agent Skills? explains the format.

How Skills Work in Cursor

Where skills live. Cursor loads skills from .agents/skills/ and .cursor/skills/ in your project, and from the same folders in your home directory for skills you want in every project. For compatibility, it also loads skills from Claude and Codex folders such as .claude/skills/ (Cursor docs: Agent Skills). Project skills are ordinary files, so they sit in your repository with everything else.

Using a skill. When Cursor starts, it finds your skills and makes them available to the agent, which decides when one is relevant. To choose one yourself, type / in Agent chat and search for its name. To keep a skill on for a whole session, use it as a Custom Mode with Option+Enter on a Mac or Alt+Enter on Windows.

Controlling when it applies. Set disable-model-invocation: true and the skill only runs when someone types its name, like a traditional slash command. Add a paths field and the skill only appears when the agent is working with matching files.

Creating a skill. Type /create-skill and describe the job. Cursor's built-in skill walks you through naming, structuring and saving it (Cursor Help: Skills). Every skill Cursor has found is listed under Customize, then Skills.

Cursor Skills vs Rules

Cursor also has rules, and the two are easy to confuse. Rules are short guidelines that Cursor includes in every conversation, or in conversations about matching files. Skills are longer, step-by-step procedures the agent uses when a job calls for them (Cursor Help: Skills).

"Use New Zealand spelling in user-facing text" is a rule. "Write the client release notes: gather the merged changes, group them and check each against its ticket" is a skill. If you already have dynamic rules or custom commands, the built-in /migrate-to-skills converts the eligible ones into skills (Cursor docs: Agent Skills).

Sharing Skills With Your Team

There are two routes. Project skills committed to your repository reach everyone who works in it, and changes go through your normal review. Personal skills in ~/.cursor/skills/ stay on your machine. On Teams and Enterprise plans, you can publish a personal skill to your team's marketplace from Customize, then Skills. Cursor packages it as a plugin, teammates choose whether to install it, and you push later changes with Sync changes (Cursor docs: Plugins).

Worked Example: Client Release Notes

Kowhai Software, a fictional software company, sends clients release notes after each release. Developers write them in Cursor, and each set reads differently. Some describe internal clean-up work clients don't care about, and one promised a feature date nobody had agreed.

A first skill for this job, saved as .cursor/skills/client-release-notes/SKILL.md in the main repository:

Description: Writes client-facing release notes from the changes merged since the last release. Use for release notes, change summaries and "what's new" updates for clients.
Inputs: the changes merged since the last release, their linked tickets and the client-facing product names.
Decisions: leave out internal-only changes. Flag any change with no ticket rather than describing it. Never give dates for future work.
Output: three sections (New, Improved, Fixed), in plain English, under 300 words.
Check before finishing: every item traces back to a merged change.

The skill holds the method. The product naming guide and the release notes template belong in the knowledge base. Cursor reaches outside tools and data through MCP servers (Cursor docs: MCP), so if those documents live in SharePoint or Google Drive, Cursor needs its own working connection to that storage. A copy pasted into the repository works too, but it won't change when the original does.

Try It: Create a Skill in Cursor

Choose one job your team repeats, with a result you can check.
In Agent chat, type /create-skill and describe the job, its inputs and the output you want.
Read the draft and add what to do when information is missing.
Tighten the description so it names the task and the words people use to ask for it.
Save it under .cursor/skills/ in the project, so it's versioned with your code.
Check it appears under Customize, then Skills, and run it by typing / and its name.
Run the three tests below before you commit it for the team.

Test It: Three Cases

A normal request. Ask for release notes for the latest release. Check every item against the merged changes, not just the tone.
Incomplete input. Include a change with no linked ticket. The notes should flag it, not invent a description.
An unrelated request. Ask the agent to fix a failing test. The release notes skill shouldn't take over. If it does, narrow the description or set disable-model-invocation: true.

Where Contexta Fits

Skills in a repository work well for the people who work in that repository. Most teams also use Claude or ChatGPT, and each tool ends up with its own copy. A skill tells your AI what to do. Your knowledge base shows it how, in detail. Your company context tells it who it's for. The Contexta harness brings them together, in every AI.

Contexta reaches Cursor through an MCP connection, and Cursor is tested with Contexta. You add Contexta as a remote MCP server using the connection URL from your dashboard, then sign in; Integrations has the setup. Each session then starts with your company context, covering terminology, naming and formats, set for the company and adjusted for each team, and the published skill for the job.

Your knowledge base stays in your own SharePoint or Google Drive. Contexta doesn't store or index your documents, so Cursor needs its own access to them. Company, team and personal spaces route each piece of work to its own folders and rules. Publish a change and connected AIs fetch it on next use, while a downloaded SKILL.md stays a separate copy. Version history lets you compare and restore. See How It Works.