How to use the Genkit JS Developer Guide agent skill

What the skill makes the agent do

The skill is a playbook for building AI-powered applications with Genkit, an open-source framework for JavaScript and TypeScript. When it is loaded, the agent checks that the Genkit CLI (genkit-cli) is installed at version 1.29.0 or newer, and it looks up current documentation with the Genkit CLI instead of relying on memory, because the source says Genkit's API changed significantly.

For conversational work, the skill tells the agent to prefer the newer beta agent API (ai.defineAgent) over a hand-written generate-plus-tools loop. It defaults to Google AI as the model provider unless you name another one. When an error occurs, it directs the agent to a list of deprecated APIs and their replacements before it attempts any fix.

• Checks the Genkit CLI version and installs or upgrades it if needed
• Uses the beta agent API for multi-turn conversations and multi-step agents
• Defaults to Google AI when no provider is specified
• Reads the official docs with the Genkit CLI rather than guessing
• Follows a fixed error procedure that starts with the Common Errors reference

Install the skill

You can install the skill in one of two ways. Download the ZIP from the skill's page on this site and unzip it into your personal skills folder, ~/.claude/skills/, or into a project's .claude/skills/ folder if you want it to apply only to that project.

If you use the Claude apps instead, open Settings, go to Capabilities, and upload the skill under Skills. The skill folder should contain SKILL.md at its top level, with the reference files alongside it in a references/ folder.

• Personal: unzip into ~/.claude/skills/
• Project only: unzip into your project's .claude/skills/
• Claude apps: Settings → Capabilities → Skills, then upload

Set up your machine for Genkit

The skill assumes you already have Node.js and npm. It also needs the Genkit CLI at version 1.29.0 or newer. The skill's own instructions say to run genkit --version to check, and to install or upgrade with npm install -g genkit-cli@^1.29.0 if the CLI is missing or older.

The beta agent API needs the genkit package at version 1.39.0 or newer. For the default model provider you need the Google AI plugin (@genkit-ai/google-genai) and a Google AI API key. If you use other providers, you need their plugins and credentials as well.

• Node.js and npm
• Genkit CLI (genkit-cli) 1.29.0 or newer
• genkit 1.39.0 or newer for the agent API
• @genkit-ai/google-genai and a Google AI API key for the default provider
• Plugins and credentials for any other provider you use

Trigger the skill with natural requests

You do not need to name the skill. The description tells the agent to use it when you ask about Genkit, AI agents, flows, or tools in JavaScript or TypeScript, or when you hit Genkit errors, validation issues, type errors, or API problems. Plain requests are enough, and including the word Genkit makes the match clearer.

Give the agent your real context, such as the framework you use or the error text. The skill mentions Next.js, Firebase and Express, and looks for @genkit-ai/next, @genkit-ai/firebase or @genkit-ai/google-cloud in package.json to detect the runtime.

• "Build a Genkit chatbot in TypeScript that remembers earlier messages."
• "Add a Genkit flow that summarises a product description using Google AI."
• "My Genkit flow throws a ValidationError when I call it. Help me fix it."
• "Why does response.text() fail in my Genkit code?"
• "Set up Genkit in this Next.js project."

Examples of what the agent produces

For a single-shot task, the agent typically writes a flow. The source's Hello World example initialises Genkit with the Google AI plugin, defines a flow with a typed input and output, and calls ai.generate with googleAI.model('gemini-flash-latest'). Review the output as you would any generated code, and run a type check such as npx tsc --noEmit, which the skill asks the agent to do after changes.

For a conversation, the agent is directed to define an agent with ai.defineAgent and to use the beta import paths, where server APIs come from genkit/beta. Check that the import paths match the docs the agent read.

What the skill needs and what it can do

The skill runs scripts and uses the network. The Genkit CLI commands it suggests, such as genkit docs:read and genkit docs:search, read documentation, and the agent may run your app through the genkit CLI to capture traces. Your model calls go to the provider you configure, which needs an API key. Read the commands the agent proposes before you approve them, and keep API keys in environment variables or secret stores rather than in source files.

The beta agent API is flagged as beta in the skill itself, so the surface may change. Pin your package versions and test before relying on it in production.

• Runs scripts: the Genkit CLI and npm commands it proposes
• Uses network: documentation lookups and calls to your model provider
• Keep provider API keys out of source control
• Review commands before approving them

Tips for better results

Tell the agent which Genkit version your project uses and whether you need the stable API or the beta agent API. If the agent reports an error, paste the full message so it can match it against the Common Errors reference, which the skill requires it to read first.

Ask the agent to verify with traces rather than a blind run, since the skill notes that running with node, tsx or npm start does not capture dev traces. Use genkit start to see traces in the Dev UI.

• State your framework (Next.js, Firebase, Express) and provider up front
• Paste the complete error message, not a paraphrase
• Ask for genkit start traces when you need to verify behaviour