Skills developing-genkit-js
📦

developing-genkit-js

Low Risk ⚙️ External commands

Build AI Applications with Genkit

Also available from: firebase

Genkit simplifies AI application development in JavaScript and TypeScript, but navigating its API changes and best practices can be time-consuming. This skill provides expert guidance on flows, tools, and providers to help you build production-ready AI applications quickly.

Supports: Claude Codex Code(CC)
🥉 72 Bronze
1

Download the skill ZIP

2

Upload in Claude

Go to Settings → Capabilities → Skills → Upload skill

3

Toggle on and start using

Test it

Using "developing-genkit-js". Create a Genkit flow that translates text between languages

Expected outcome:

Created flow 'translateText' with input schema (text, sourceLang, targetLang) and output schema (translatedText). Uses googleAI model 'gemini-2.5-flash'. Flow is registered and ready for testing with `genkit flow:run translateText`.

Using "developing-genkit-js". Help me fix this error: 'response.text is not a function'

Expected outcome:

This error indicates you are using pre-1.0 Genkit syntax. In Genkit v1.x, response properties are accessed directly, not as methods. Change `response.text()` to `response.text` in your code. See references/common-errors.md for additional migration notes.

Using "developing-genkit-js". Set up Genkit in my existing Next.js project

Expected outcome:

Detected Next.js project structure. Installing @genkit-ai/next and @genkit-ai/google-genai packages. Creating src/ai directory with genkit.ts configuration. Adding genkit:ui script to package.json. Genkit is now configured and ready for use.

Security Audit

Low Risk
v1 • 4/19/2026

Static analysis flagged 194 potential security issues, primarily CLI command examples in markdown documentation. All flagged patterns are documentation showing Genkit CLI usage (genkit start, npm install, genkit docs:search, etc.). No actual executable code patterns or malicious intent found. The skill provides standard Genkit development patterns for AI application development.

6
Files scanned
543
Lines analyzed
5
findings
1
Total audits
Medium Risk Issues (1)
CLI Command Examples in Documentation
176 markdown code blocks show CLI commands (npm install, genkit start, etc.). These are documentation examples, not executable code. No user input injection vectors exist.
Low Risk Issues (3)
Hardcoded Localhost URL
Documentation references http://localhost:4000 for local development server. Standard development pattern, not a security risk.
Example URL in Documentation
Documentation uses https://example.com/photo.jpg as placeholder image URL. Standard example domain, not sensitive.
API Key Reference in Documentation
Documentation mentions GEMINI_API_KEY environment variable. Reminds users to configure their own keys, does not contain or transmit secrets.
Audited by: claude

Quality Score

45
Architecture
100
Maintainability
87
Content
50
Community
79
Security
91
Spec Compliance

What You Can Build

New Project Setup

Initialize a new Genkit project with proper structure, configure a provider, and create a simple working flow. Ideal for developers getting started with AI application development.

Error Resolution

Debug Genkit validation errors, type mismatches, and deprecated API usage. The skill includes a comprehensive common errors reference with solutions.

Advanced Flow Design

Build complex multi-step flows with custom tools, streaming responses, and multimodal capabilities (text, image, audio generation).

Try These Prompts

Create a Simple Flow
Create a Genkit flow that takes a topic as input and returns a summary. Use the Google AI provider with gemini-2.5-flash model.
Build Multi-Tool Agent
Create a Genkit flow with two tools: one to search the web and another to format results. Chain them together in a single flow.
Implement Streaming
Convert a regular Genkit flow to use streaming output so users see responses as they are generated.
Debug Genkit Error
I am getting a 'configureGenkit is not a function' error when running my Genkit code. The code worked yesterday but I updated some packages. Help me fix this.

Best Practices

  • Always use the latest model versions (gemini-2.5-flash or newer) and consult the Genkit CLI for current recommendations
  • Use Zod schemas with .describe() for all input and output fields to guide the LLM behavior
  • Run type checks (npx tsc --noEmit) after writing Genkit code to catch schema mismatches early

Avoid

  • Do not use deprecated model versions (gemini-1.5-*, gemini-2.0-*) which are decommissioned and will fail
  • Do not import from internal packages (@genkit-ai/ai, @genkit-ai/core, @genkit-ai/flow) - import from the main 'genkit' package
  • Do not rely on internal knowledge for error fixes - always consult references/common-errors.md first before other sources

Frequently Asked Questions

Which AI providers does this skill support?
Genkit supports multiple providers including Google AI (default), OpenAI, Anthropic, Ollama, and Vertex AI. Search docs with 'genkit docs:search plugins' to find specific packages.
How do I update the Genkit CLI?
Run 'npm install -g genkit-cli@^1.29.0' to get the latest version. Verify with 'genkit --version'.
Why am I getting type errors with my Genkit code?
Most type errors come from deprecated imports. Ensure you import { z, genkit } from 'genkit' and plugins from their specific packages like '@genkit-ai/google-genai'.
How do I test my flows locally?
Run 'genkit start -- <your command>' to start the Developer UI. Use 'genkit flow:run <flowName>' for CLI-based testing.
Can I use Claude or OpenAI models with Genkit?
Yes. Use the @genkit-ai/oai-compat plugin for OpenAI models or search for Anthropic-specific plugins using 'genkit docs:search anthropic'.
What is the difference between flows and tools?
Flows are multi-step AI operations that can call multiple tools and return complex results. Tools are single-purpose functions that can be called by flows or agents.

Developer Details