Repository Public
1AI coding assistants are powerful but have limitations---training data cuts off2at a specific date, missing new API features and changes. Without access to3Gemini-specific documentation, agents may suggest generic patterns instead of4optimized approaches.56To keep your coding assistant current with the evolving Gemini API and its7recommended usage, we recommend setting up the **Gemini Docs MCP**8and enhancing your environment with **Gemini API Skills**. While these tools are usable independently, they are designed to work9together to provide complete coverage.1011## Connect the Gemini Docs MCP1213Gemini hosts a public Model Context Protocol (MCP) server at14`https://gemini-api-docs-mcp.dev`. Connecting your coding agent to this server ensures that15all queries have access to the latest APIs, code updates, and optimal configuration examples.1617Run the following command in your agent's terminal or project root to install18the server:1920 npx add-mcp "https://gemini-api-docs-mcp.dev"2122This server adds a `search_documentation` function that your agent can use to23retrieve real-time API definitions and integration patterns from the official24Gemini documentation files.2526## Add API Development Skills2728The skills provide **baked-in rules and best practices** (such as enforcing the correct SDK and current model versions) directly in your assistant's context. The skill works together with the Gemini Docs MCP service: If you have both29installed, the skill uses the MCP service for documentation, but even without30the MCP installed, it will fetch `llms.txt` from `ai.google.dev` as a fallback.3132To install these skills, you can use one of the following supported tools. Installation instructions for both are provided below each skill module:3334- **[skills.sh](https://skills.sh)**: Recommended. The open standard for portable agent behaviors.35- **[Context7](https://context7.com)**: Supported for users already utilizing the Context7 ecosystem.3637### gemini-api-dev3839The foundational skill for general-purpose Gemini development. This skill provides documentation and best practices for:4041- Prompt routing to current models (e.g., Gemini 3.1 Pro/Flash) and avoiding deprecated models42- Multimodal prompting, function calling, structured outputs, and common integration patterns4344#### Install with skills.sh4546 npx skills add google-gemini/gemini-skills --skill gemini-api-dev --global4748#### Install with Context74950 npx ctx7 skills install /google-gemini/gemini-skills gemini-api-dev5152### gemini-live-api-dev5354Skill for building real-time conversational AI applications with Gemini Live55API. This skill provides documentation and best practices for:5657- WebSocket connections for low-latency streaming58- Streaming audio, video, and text59- Voice activity detection and barge-in support6061#### Install with skills.sh6263 npx skills add google-gemini/gemini-skills --skill gemini-live-api-dev --global6465#### Install with Context76667 npx ctx7 skills install /google-gemini/gemini-skills gemini-live-api-dev6869### gemini-interactions-api7071Skill for building apps with the72[Interactions API](https://ai.google.dev/gemini-api/docs/interactions-overview). The Interactions API is our simplest and best way to build with Gemini models and agents. This skill covers:7374- Text generation, multi-turn chat, and streaming75- Function calling, structured output, and image generation76- Background execution and Deep Research agents77- Server-side conversation state management78- Python and TypeScript SDK patterns7980#### Install with skills.sh8182 npx skills add google-gemini/gemini-skills --skill gemini-interactions-api --global8384#### Install with Context78586 npx ctx7 skills install /google-gemini/gemini-skills gemini-interactions-api8788## Verify installation8990After installing, confirm that your coding assistant can connect to the Gemini Docs MCP server and use your installed skills.9192### 1. Verify agent behavior9394The most reliable way to verify is to ask your agent a technical question about Gemini API.9596**Prompt:** "How do I use context caching with the Gemini API?"9798A successful setup will:99100- **Provide accurate code** : Reference specific Gemini methods like `cacheContent` or `cachedContents.create` from the latest endpoints.101- **Use the MCP Tool** : Show that it is connected to the **Gemini Docs MCP Server** or utilizing the `search_documentation` tool to fetch data.102- **Invoke loaded skills**: Show an indicator that it is "Using skill: gemini-api-dev" (if relying on a secondary wrapper).103104### 2. Verify manifestations \& tools105106If the agent gives a general or generic answer, use the specific Discovery or Status commands for your environment to verify that the Docs MCP or skill is loaded into memory.107108| Environment | MCP Verification | Skills Verification |109|---|---|---|110| **Claude Code** | Type `/mcp` in the terminal to view active servers and `search_documentation` tools. | Type `/skills` in the terminal to list all active manifests. |111| **Cursor** | Navigate to **Settings \> Features \> MCP**. Ensure server is "Connected". | Open **Settings \> Rules**. Verify the skill appears under "Agent Decides." |112| **Antigravity** | Check the **Customizations \> Connections** sidebar for MCP status. | Type `/skills list` or check the **Customizations \> Rules** sidebar. |113| **Gemini CLI** | Run `gemini mcp list` or use `/mcp list`. | Run `gemini skills list` or use the `/skills` slash command in-session. |114| **Copilot** | Type `@gemini /mcp` to list active data connectors. | Type `@gemini /skills` (or `/skills`) to view active extensions. |115116## Troubleshooting117118If your agent provides only general information or fails to recognize119Gemini-specific methods, check the following:120121### Agent didn't discover the skill122123Most agents index skills only on startup.124125**Fix:** Completely restart your IDE (Cursor/VS Code) or exit and re-open your126terminal-based agent (Claude Code).127128### Global vs. local conflict129130If you installed with the `--global` flag, your agent might be ignoring it in131favor of project-specific rules.132133**Fix:** Try installing the skill directly into your project root without the134global flag:135136 npx skills add google-gemini/gemini-skills --skill gemini-api-dev137138## Resources139140- [Gemini API skills on GitHub](https://github.com/google-gemini/gemini-skills)141- [Interactions API](https://ai.google.dev/gemini-api/docs/interactions-overview)142- [Get started](https://ai.google.dev/gemini-api/docs/get-started)143- [Libraries](https://ai.google.dev/gemini-api/docs/libraries)