ai-sdk-agents
Expert guidance for building AI agents with ToolLoopAgent (AI SDK v6+). Use when creating agents, configuring stopWhen/prepareStep, callOptionsSchema/prepareCall, dynamic tool selection, tool loops, or agent workflows (sequential, routing, evaluator-optimizer, orchestrator-worker). Triggers: ToolLoopAgent, agent loop, stopWhen, stepCountIs, prepareStep, callOptionsSchema, prepareCall, hasToolCall, InferAgentUIMessage, agent workflows.
Installation
Details
Usage
After installing, this skill will be available to your AI coding assistant.
Verify installation:
npx agent-skills-cli listSkill Instructions
name: ai-sdk-agents description: | Expert guidance for building AI agents with ToolLoopAgent (AI SDK v6+). Use when creating agents, configuring stopWhen/prepareStep, callOptionsSchema/prepareCall, dynamic tool selection, tool loops, or agent workflows (sequential, routing, evaluator-optimizer, orchestrator-worker). Triggers: ToolLoopAgent, agent loop, stopWhen, stepCountIs, prepareStep, callOptionsSchema, prepareCall, hasToolCall, InferAgentUIMessage, agent workflows.
AI SDK Agents
Build autonomous agents with ToolLoopAgent: reusable model + tools + loop control.
Quick Start
Assume Zod v4.3.5 for schema typing.
import { ToolLoopAgent, tool } from 'ai';
import { anthropic } from '@ai-sdk/anthropic';
import { z } from 'zod';
const weatherAgent = new ToolLoopAgent({
model: anthropic('claude-sonnet-4-20250514'),
tools: {
weather: tool({
description: 'Get the weather in a location (F)',
inputSchema: z.object({ location: z.string() }),
execute: async ({ location }) => ({ location, temperature: 72 }),
}),
},
});
const result = await weatherAgent.generate({
prompt: 'What is the weather in San Francisco?',
});
When to Use ToolLoopAgent vs Core Functions
- Use ToolLoopAgent for dynamic, multi-step tasks where the model decides which tools to call.
- Use generateText/streamText for deterministic flows or strict ordering.
Essential Patterns
Structured Output
import { ToolLoopAgent, Output } from 'ai';
import { z } from 'zod';
const analysisAgent = new ToolLoopAgent({
model: 'openai/gpt-4o',
output: Output.object({
schema: z.object({
sentiment: z.enum(['positive', 'neutral', 'negative']),
summary: z.string(),
}),
}),
});
Streaming Agent
const stream = myAgent.stream({ prompt: 'Summarize this report' });
for await (const chunk of stream.textStream) {
process.stdout.write(chunk);
}
API Route
import { createAgentUIStreamResponse } from 'ai';
export async function POST(request: Request) {
const { messages } = await request.json();
return createAgentUIStreamResponse({ agent: myAgent, messages });
}
Type-Safe Client Integration
import { ToolLoopAgent, InferAgentUIMessage } from 'ai';
const myAgent = new ToolLoopAgent({ model, tools });
export type MyAgentUIMessage = InferAgentUIMessage<typeof myAgent>;
Loop Control Checklist
- Set
stopWhen(default:stepCountIs(20)) for safety. - Use
hasToolCall('finalAnswer')to stop on terminal actions. - Use
prepareStepto swap models, compress messages, or limit tools per step.
Runtime Configuration
- Use
callOptionsSchemato define type-safe runtime options. - Use
prepareCallto select model/tools or inject RAG context once per call. - Use
prepareStepfor per-step decisions (budget limits, dynamic tools).
Reference Files
| Reference | When to Use |
|---|---|
references/fundamentals.md | ToolLoopAgent basics, Output types, streaming |
references/loop-control.md | stopWhen, hasToolCall, prepareStep patterns |
references/configuration.md | callOptionsSchema, prepareCall vs prepareStep |
references/workflow-patterns.md | multi-agent workflows and routing |
references/real-world.md | RAG, multimodal, file processing |
references/production.md | monitoring, safety, cost control |
references/migration.md | v6 migration notes |
More by BjornMelin
View allExpert guidance for building chat UIs with AI SDK React hooks. Use when: (1) Building chatbots with useChat hook, (2) Implementing tool UIs with server/client execution, (3) Handling message persistence and stream resumption, (4) Creating generative UI with React components, (5) Integrating with Next.js/Node/Fastify/Nest backends, (6) Using useObject for streaming structured data. Triggers: useChat, useObject, useCompletion, chat UI, chatbot, generative UI, message persistence, @ai-sdk/react, UIMessage, tool parts, streaming UI, toUIMessageStreamResponse.
Production-ready Supabase integration patterns for Next.js/React/TypeScript applications. Use when working with Supabase for (1) SSR authentication with @supabase/ssr, (2) Database operations and migrations, (3) Row Level Security (RLS) policies, (4) Storage buckets and file uploads, (5) Realtime channels and presence, (6) Edge Functions with Deno, (7) pgvector embeddings and semantic search, (8) Vercel deployment and connection pooling, (9) CLI operations (type generation, migrations). Triggers on Supabase client setup, auth patterns, RLS policies, storage uploads, realtime subscriptions, Edge Functions, vector search, or Vercel+Supabase deployment.
Expert guidance for building Dash applications with Dash Mantine Components (DMC) v2.x. Use when creating dashboards, forms, data visualization apps with DMC. Covers: MantineProvider theming, style props (m, p, c, bg, w, h), Styles API, callbacks (basic, pattern-matching ALL/MATCH/ALLSMALLER, clientside, background), multi-page apps with Dash Pages, charts (LineChart, BarChart, DonutChart), date pickers, modals, and all 100+ components. Triggers on: dash-mantine-components, DMC, MantineProvider, dmc.Button, dmc.Select, dmc.Modal, dmc.BarChart, Mantine theme, Dash UI components, Dash callbacks, multi-page Dash app, pattern-matching callbacks, clientside callbacks, AppShell.
World-class Vitest QA/test engineer for TypeScript + Next.js (local + CI performance focused)