typescript is published by Nathan Shively-Sanders, Ryan Cavanaugh, Emily Fischer from microsoft/typescript-skills, pinned to 2480764651ce9afbb7b29c56bb58e9a16b80e7b9. Review the source before installing.
Skills for coding agents working with TypeScript, JavaScript, and Node.js. The marketplace and plugin layout follows dotnet/skills. Skills use the Agent Skills format.
What's included
The typescript plugin currently includes one skill:
| Skill | Purpose |
|---|---|
| typescript-project-setup | Create a new TypeScript project with a suitable configuration. |
The skill provides instructions for installing TypeScript, configuring a new project, and setting up builds. Automatic invocation is not disabled.
Installation
The following commands install the plugin and its project setup skill.
Copilot CLI and Claude Code
Add the marketplace, then install the plugin:
/plugin marketplace add microsoft/typescript-skills
/plugin install typescript@typescript-skills
Restart the host if needed. Update the plugin manually with:
/plugin update typescript@typescript-skills
VS Code
For VS Code versions supporting agent plugins, enable plugin discovery in user settings:
{
"chat.plugins.enabled": true,
"chat.plugins.marketplaces": ["microsoft/typescript-skills"]
}
Use /plugins in Copilot Chat to browse the marketplace and install typescript.
Codex CLI
For Codex versions supporting plugins:
codex plugin marketplace add microsoft/typescript-skills
Open /plugins and install typescript from typescript-skills.
The Codex marketplace manifest is .agents/plugins/marketplace.json.
Cursor
The repository includes a Cursor marketplace manifest, but this does not mean it is published in Cursor's hosted marketplace. For local development, use Cursor's local plugin import mechanism with this checkout.
Development
Use Node.js 24 or later:
npm ci
npm run validate
This runs structural plugin/skill checks, checker regression tests, and Markdown linting. It does not execute or evaluate the skills in an agent host.
See CONTRIBUTING.md for authoring and manual versioning, and tests/README.md for the initial benchmark results. The results describe upstream experiments, not evaluations of the standalone skill distributed in this repository. There is no dashboard, scheduled agent evaluation, or automatic release process.
License
MIT.