gemini-api-dev Skill for Claude: Write Current Gemini API Code

Gives Claude the current Gemini models, SDKs and Interactions API rules so the code it writes for the Gemini API is up to date.

Installs: 25.8K; on skills.sh
Stars: 4.4K; on GitHub
Files: 1; besides SKILL.md
Updated: Oct 2026; last repo commit

gemini-api-dev corrects what Claude assumes about the Gemini API. SKILL.md says its rules override Claude's training data. It lists the current models and agents, the SDK versions to use, and the Interactions API call shape. It tells Claude to avoid the legacy SDKs and old model families, and to fetch the matching hosted doc page before it writes code.

It triggers when you write code that calls Gemini for text, chat, images, video, speech, streaming, function calling, structured output or background research tasks. For migration work, it reads references/migration.md and asks you to confirm scope before it edits anything.

Good for

Developers building apps on the Gemini API in Python or TypeScript
Teams moving old generateContent code or retired models to the Interactions API
Anyone building chat, image, speech, video, research or agent features with Gemini

Install gemini-api-dev

Claude Code

Copy this prompt.

Install the gemini-api-dev skill in Claude Code. Run this command:

npx skills add https://github.com/google-gemini/gemini-skills --skill gemini-api-dev

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

From its listing on skills.sh. Source.

What's in the skill

When it loads

Use this skill when writing code that calls the Gemini API for text generation, multi-turn chat, multimodal understanding, image generation, video generation, speech generation (TTS), voice design, voice replication, streaming responses, background research tasks, function calling, structured output, or migrating from the old generateContent API. Covers SDK usage and best practices for Gemini models and agents in Python and TypeScript.

Sections in SKILL.md

Critical Rules (Always Apply)
Current Models (Use These)
Current Agents
Current SDKs
Important Additional Notes
Quick Start
Python
JavaScript/TypeScript
Response Helpers
Stateful Conversation
Python
JavaScript/TypeScript
Deep Research Agent
Python
JavaScript/TypeScript
Managed Agents

Key facts

Author: Google Gemini team repo (google-gemini); the README says it is not an officially supported Google product
License: Apache-2.0 (as reported by GitHub)
What installs: 2 files including SKILL.md (references/migration.md); no scripts
SDKs it names: Python google-genai 2.30.1 or newer; JavaScript @google/genai 2.29.1 or newer
Needs: A Gemini API key for the code it writes; it also fetches docs from aistudio.google.com
Vendor claim: README: adding the skill raised correct API code to 87% with Gemini 3 Flash and 96% with Gemini 3.1 Pro
Installs on skills.sh: 25,842 (Nov 2, 2026)
GitHub stars: 4,367 (Nov 2, 2026)
Last commit: 2026-10-30
skills.sh audits: Gen Agent Trust Hub pass, Socket pass, Snyk pass

What it includes

Lists the current models, agents and SDKs and bans legacy ones
SKILL.md says its model list overrides what Claude knows, and that a request for a deprecated model should be swapped to gemini-3.8-flash with a note. We have not verified those model IDs.
Critical rules
Reads the matching aistudio.google.com page before writing code
SKILL.md says the hosted pages are the source of truth for parameters and edge cases, and lists pages from text generation to Deep Research.
Network · Fetch the docs first
Quick starts, stateful chat, streaming and response helpers
Shows Python and TypeScript for client.interactions.create, previous_interaction_id for follow-ups, and stream events.
Interactions API basics
Runs a long research task in the background and polls for the result
Uses a deep-research agent with background=True and checks the status every 10 seconds.
Network · Deep Research agent
Runs an agent in a Google-hosted Linux sandbox
The Antigravity agent can run Bash, Python and Node.js, manage files and browse the web. Custom agents can be created from a base agent.
Runs code · Managed agents
Step types, content types, stream events and status values
A cheat sheet for reading Interaction responses.
Data model reference
Moves old generateContent code and retired models to current ones
Starts with a scope check, then a checklist and before and after code, including TTS and image-model upgrades.
Write · references/migration.md
Sends you to gemini-live-api-dev for real-time audio and video
SKILL.md says to install the google-gemini/gemini-live-api-dev skill for real-time streaming.
Live API pointer

From the official docs

Permissions

Medium risk

It leads Claude to install Google's SDKs, call the Gemini API with your key and, if you use managed agents, run code in a Google-hosted sandbox. Use a key with a spending limit and keep it out of source files.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

Interactions are stored by default
SKILL.md says paid tier keeps stored interactions for 55 days and free tier for 1 day. Set store to false to opt out, which also turns off previous_interaction_id and background runs. github.com
Managed agents execute code in a remote sandbox
The Antigravity agent can run Bash, Python and Node.js, manage files and browse the web. Do not give it secrets or private repositories you cannot share. github.com
It installs SDK packages
It tells Claude to use pip install -U google-genai or npm install @google/genai. Review the commands before you allow them. github.com
Migration edits existing code
references/migration.md says to confirm scope with you before editing. Work on a branch. github.com

Technical details

Talks to	aistudio.google.com documentation pages. The Gemini API, from the code it helps you write. Google-hosted sandbox for managed agents
Sign-in and keys	SKILL.md links an API Keys doc and shows genai.Client() with no key in code. Keep the key in an environment variable.
Scopes	Edits your source files when you ask for new code or a migration pip install -U google-genai or npm install @google/genai Gemini API key for the code it writes

Questions

What does gemini-api-dev do?

It makes Claude write current Gemini API code. It gives Claude the models, SDKs and Interactions API patterns it should use, and tells it to read the hosted docs before coding.

How do I install gemini-api-dev?

Use the install box above, which is the README's skills CLI route for this one skill. The README also documents plugin routes for Claude Code, Cursor and Codex that bundle the repo's other skills.

Does it need a Gemini API key?

The skill itself needs none. The code it writes calls the Gemini API, so you need a key from Google AI Studio, and the SKILL.md links an API keys page.

Which models does it tell Claude to use?

SKILL.md lists the model IDs it wants used: Gemini text, image, speech, video, embedding and Gemma models, with gemini-3.8-flash in its quick start. It calls the gemini-2.5, 2.0 and 1.5 families legacy and says to use gemini-3.8-flash instead, noting the swap. These are the skill's claims; we have not checked the IDs against Google's model list, so confirm them in the docs before you rely on them.

Does it help me migrate from generateContent?

Yes. It reads references/migration.md, confirms scope with you, then follows a checklist for the move to the Interactions API and for model strings.

Is it safe?

It is text guidance plus one reference file, but it leads Claude to fetch web docs, install SDK packages and make API calls with your key. Managed agents run code in a Google sandbox, and interactions are stored by default, so read those settings first.

Is it an official Google skill?

It lives in the google-gemini GitHub organization. Its README says it is not an officially supported Google product.

What about the Gemini Live API?

This skill sends you to the separate gemini-live-api-dev skill for real-time audio, video and text streaming. Install that one if you need it.

Information

updated Nov 2, 2026

Repository: google-gemini/gemini-skills
Folder: skills/gemini-api-dev

google-gemini/gemini-skills skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 3