google-gemini/gemini-skills

Overview

Language models stop learning at their training cutoff, so they can lag behind changes in the Gemini API and SDK. This repository packages agent skills for the Gemini API: lightweight bundles of guidance that coding assistants load to write more accurate, up-to-date Gemini code README. Skills cover the Interactions API (text, chat, streaming, function calling, structured output, image and speech generation, managed agents, Deep Research), shell access through the gemini-api CLI, real-time apps with the Live API, and video workflows with Gemini Omni 1.1 Flash skill list.

The intended audience is developers already using an AI coding assistant and building Gemini-powered applications, either in Python or TypeScript. Installation guidance is provided for several assistants, so the repository is not tied to one editor.

According to the project's own evaluation, adding the skill raised an agent's rate of generating correct API code that follows best practices to 87% with Gemini 3 Flash and 96% with Gemini 3.1 Pro performance note. These are the authors' measurements, not an independent benchmark.

Quick start

You need Node.js/npm for the CLI-based paths, or an assistant that supports plugin installation directly.

1. Browse and install skills with the Vercel skills CLI:

   npx skills add google-gemini/gemini-skills --list
   npx skills add google-gemini/gemini-skills --skill gemini-api-dev
   

   The first command lists available skills; the second installs gemini-api-dev. The installation section documents this flow.

2. Alternatively, use the Context7 CLI: npx ctx7 skills install /google-gemini/gemini-skills gemini-api-dev Context7 instructions.

3. In Claude Code, register the marketplace with /plugin marketplace add google-gemini/gemini-skills, then install with /plugin install gemini-skills@gemini-skills Claude Code setup. Cursor users can run /add-plugin google-gemini/gemini-skills Cursor setup. Codex users add the marketplace and then install gemini-skills@gemini-skills Codex setup. Antigravity bundles the skills under Customizations → Build with Google Plugins, or you can run agy plugin install https://github.com/google-gemini/gemini-skills Antigravity setup.

After installation, the assistant can consult the skill while writing Gemini API code. No separate runtime is required for the skill content itself beyond the assistant and its plugin mechanism.

Scope and constraints

• The skills target the Gemini API and its SDKs; they are not general-purpose Python or TypeScript instructions.
• The vertex-ai-api-dev skill has moved to the google/skills repository, so users following older references should install it from there migration note.
• This is not an officially supported Google product and is not eligible for the Google Open Source Software Vulnerability Rewards Program disclaimer.
• The repository is licensed under Apache-2.0, and additional setup guidance for coding agents is available in the Gemini API docs.

HotRepo's take

On the Python daily list for 2026-11-05, the project recorded 4,396 stars and 481 forks, with 12 stars gained that day and a best position of 14 snapshot. That is a single recorded snapshot, so it establishes the day's placement but not a broader growth trend. The repository's own evaluation separates it from generic prompt collections: the value lies in concrete, version-sensitive API guidance for Gemini, and its spread across multiple assistant installers suggests it is meant to be used directly rather than read as a static document.