Gemini CLI Skills for AI Coding: What Claude Code Users Need to Know (2026)
Google's Gemini CLI enters the AI coding space as Claude Code's closest competitor. The Skillselion catalog has 83,206 listings including `google-agents-cli-publish` (27,779 installs, 2,729 stars) and `gemini-api-dev` (14,410 installs, 3,606 stars)-covering Gemini API integration, agent publishing, and Firebase AI Logic. Here's how the ecosystems compare and where skills bridge them.
By Skillselion, an Ellelion LLC publication · Updated July 5, 2026 · 3 min read · Stats verified against the live catalog
Google launched Gemini CLI in mid-2025 as a direct Claude Code competitor-a terminal-based AI coding assistant with a 2M-token context window and Google Search grounding. The Skillselion catalog reflects the market reality: Claude Code skills dominate with 83,206 listings, but Gemini API skills are climbing fast, and Google's agent-publishing workflow has its own high-install skill cluster. Here's how both ecosystems sit today and what Claude Code users need to know.
Key takeaways
firebase-ai-logic-basicstops Gemini-ecosystem installs at 50,505 installs and 345 stars (skills.sh registry, GitHub)-Firebase's AI backend SDK for Gemini model integration in Firebase Functions.google-agents-cli-publish(27,779 installs, 2,729 stars) covers publishing Claude-Code-like agents to Google's Agent Garden-a workflow for distributing AI tools across the Google ecosystem.gemini-api-dev(14,410 installs, 3,606 stars) teaches calling the Gemini API directly:generateContent, multimodal inputs, and streaming-useful for building Gemini-backed features while authoring in Claude Code.- Claude Code's skill catalog (83,206 listings) has a multi-year head start over Gemini CLI's ecosystem-the skill breadth advantage compounds with every new MCP server published.
- Gemini CLI's 2M context window is its differentiator for whole-repository analysis; Claude Code's skill system is its differentiator for repeatable, community-vetted coding patterns.
What is Gemini CLI and how does it compare to Claude Code?
Gemini CLI is Google's terminal AI coding tool, launched in mid-2025. It uses Gemini 2.5 Pro/Flash models, supports MCP servers, and has native Google Search grounding and Google Workspace integration. The key architectural difference: Gemini CLI's context window (2M tokens) can ingest entire large repositories in one pass; Claude Code's skill system provides reusable, community-vetted patterns that accumulate across sessions and projects.
What are the best Gemini skills in the catalog?
firebase-ai-logic-basics (50,505 installs, 345 stars, skills.sh registry) is the top Gemini-ecosystem skill by install count-it covers Firebase's AI Logic SDK, which embeds Gemini model calls into Firebase Functions, Firestore triggers, and App Hosting. If you're building a Firebase backend with AI features, this is the first install.
claude skill install gemini-api-dev
claude skill install google-agents-cli-publishgemini-api-dev (14,410 installs, 3,606 stars) handles direct Gemini REST/SDK calls; google-agents-cli-publish (27,779 installs, 2,729 stars) covers agent distribution.
Can you use Gemini API skills inside Claude Code?
Yes-gemini-api-dev (14,410 installs, 3,606 stars) teaches Claude Code to generate Gemini API integration code: GoogleGenerativeAI SDK setup, generateContent calls with multimodal inputs (text + images + PDFs), streaming with generateContentStream, and safety-setting configuration. This is useful for building product features that use Gemini as the inference backend while the development workflow stays in Claude Code. See the MCP server build guide for wrapping Gemini as an MCP tool.
What is Google Agents CLI and why does it matter?
google-agents-cli-publish (27,779 installs, 2,729 stars) teaches the workflow for publishing AI agents to Google's Agent Garden-the distribution platform for Gemini CLI extensions. For Claude Code users, this skill is valuable if you're building cross-platform AI tools that need to work in both Claude Code and Gemini CLI environments: it covers the shared agent.json manifest format and tool-call contract that both runtimes respect.
How do Gemini skills pair with MCP in Claude Code?
Both Claude Code and Gemini CLI support MCP servers-the tool protocol is the common layer. A Gemini-backed MCP server built with gemini-api-dev skill patterns is callable from Claude Code via the standard MCP client. See the MCP server guide and best MCP server skills for the build-and-publish workflow. The security skills guide covers API key management and rate-limit handling for Gemini API integrations.
A complete starter stack for Gemini integration
gemini-api-dev- REST/SDK calls, multimodal inputs, streamingfirebase-ai-logic-basics- Firebase Functions + Gemini AI Logic backendgoogle-agents-cli-publish- agent manifest, Agent Garden distributionfirebase-ai-logic(34,077 installs) - advanced Firebase AI Logic patterns
Common pitfalls
- Using
gemini-api-devwithout configuring safety thresholds-Gemini models have four harm categories that default to BLOCK_MEDIUM_AND_ABOVE; production code typically needs explicitBLOCK_ONLY_HIGHsettings for content-generation features. - Installing
firebase-ai-logic-basicsexpecting Gemini CLI tooling-this skill covers the Firebase AI backend SDK, not the Gemini CLI terminal tool itself. - Building a Gemini MCP server without input validation-Claude Code's security skills cover the injection attack surface that's specific to LLM tool calls.
Explore the full MCP server ecosystem to see how Gemini, Claude, and other model APIs fit into a unified tool-call architecture.
Common questions
What is Gemini CLI and how does it compare to Claude Code?
Gemini CLI is Google's terminal-based AI coding assistant, analogous to Claude Code. It uses Gemini models and integrates with Google Cloud and Firebase. The key difference: Claude Code has a richer skill ecosystem (83,206 listings vs Gemini CLI's nascent tooling), but Gemini CLI's 2M context window handles larger codebases in a single pass.
What Gemini skills are available in the Skillselion catalog?
The catalog has three high-volume Gemini skills: `firebase-ai-logic-basics` (50,505 installs, 345 stars), `google-agents-cli-publish` (27,779 installs, 2,729 stars), and `gemini-api-dev` (14,410 installs, 3,606 stars). All are installable in Claude Code to work with Gemini APIs from within the Claude Code workflow.
Can I use Gemini API skills with Claude Code?
Yes. `gemini-api-dev` (14,410 installs, 3,606 stars) teaches Claude Code to call the Gemini API: `generateContent`, multimodal inputs, streaming responses, and safety settings. It's useful for building features that use Gemini as a backend model while authoring the code in Claude Code. See [MCP server guide](/guide/mcp-server-build-guide-2026) for building Gemini-backed MCP tools.
What is firebase-ai-logic and why does it have so many installs?
`firebase-ai-logic-basics` (50,505 installs, 345 stars) is Firebase's AI-integrated backend SDK, formerly Firebase ML. The high install count reflects Firebase's massive developer base using Gemini models embedded in Firebase Functions and Vertex AI integrations-it's a backend AI skill, not a CLI tool.
Which AI coding tool has more skills in the catalog?
Claude Code's skill catalog has 83,206 listings vs Gemini CLI's early-stage tooling ecosystem. By install volume, Claude Code skills dominate: the top skill `remotion-best-practices` alone has 382,543 installs. Gemini CLI is growing but Claude Code has a multi-year head start in community-built skills.
Ranked by Skillselion - an independent directory of AI-coding tools, not affiliated with Anthropic, OpenAI or Cursor. Tool rankings reflect real adoption (installs, then GitHub stars) from the skills.sh registry and GitHub, last updated July 5, 2026.