Build provider-agnostic AI apps with streaming text, tool calls, and structured outputs.
npx skills add https://github.com/janpeterd/dotfiles --skill tanstack-ai-janpeterd
Please help me install this Agent Skill. Skill: tanstack-ai Source: https://github.com/janpeterd/dotfiles/tree/main/dot_agents/skills/tanstack-ai Command: npx skills add https://github.com/janpeterd/dotfiles --skill tanstack-ai-janpeterd
TanStack AI helps developers avoid brittle, provider-specific AI integrations by providing a unified, type-safe SDK for building reliable chat, streaming generation, and structured outputs.
Use TanStack AI to generate streaming, type-safe responses by creating a chat/completion flow with an adapter like openaiText and passing your messages and (optionally) a schema for structured output.
High-intent search queries and answers about installing and using this skill.
To build type-safe AI chat with streaming, you use a provider-agnostic adapter to generate streaming tokens and chunks for responsive UIs. This approach ensures typed message handling while progressively rendering text as it arrives.
Yes, you can use Zod schemas for structured output generation by converting them to JSON schema. This allows your AI application to reliably return typed objects instead of free-form text, ensuring type safety in downstream logic.
This AI SDK supports tool calling and agentic workflows by allowing you to define tools with parameters and execute them iteratively. You can implement agent loops with max-iteration limits, optional approval, and observability hooks for monitoring.
The best way to switch between OpenAI, Anthropic, and Gemini is using provider-agnostic adapters. This abstraction lets you swap common model providers without rewriting your application logic, maintaining consistent typed message handling across environments.
You handle multimodal requests by passing image inputs through the provider-agnostic adapters. The SDK supports multimodal requests across common model providers, including image generation adapters, enabling diverse media processing within your workflows.
Provider-specific AI integrations break when switching models due to brittle, provider-specific code. Using a unified, type-safe SDK prevents this by providing provider-agnostic adapters that standardize streaming, structured outputs, and tool calling.
api7Configure APISIX key-auth plugin via a6 CLI for API key authentication.
quanngynxCreate AI chat interface components for Next.js and shadcn/ui projects.
allthingslinuxDesign Zod schemas for validation and TypeScript type inference.
memohaiAbstract provider-specific AI APIs into a unified Go SDK.
