effect-uai

Builds AI agent loops with Effect-uai primitives like loop and streamTurn.

30|4|Updated Apr 26, 2026
One-click install
npx skills add https://github.com/betalyra/effect-uai --skill effect-uai
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: effect-uai
Source: https://github.com/betalyra/effect-uai/tree/main/skills/effect-uai
Command: npx skills add https://github.com/betalyra/effect-uai --skill effect-uai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Enables building AI agent loops using Effect-uai primitives without enforcing a fixed runtime, giving you ownership of state and flow while the system handles wire formats, streaming events, and provider integration.

Core Features & Use Cases

  • Streaming turns and typed events through a single, flexible loop.
  • Tool integration and structured output with typed schemas and outputs.
  • Provider-agnostic design with pluggable language model layers for OpenAI, Anthropic, Gemini, and others.
  • Reusable recipe catalog patterns to compose end-to-end agent workflows.

Quick Start

Install the core + a provider package, then scaffold a minimal loop using streamTurn to observe streaming turns and tool outputs.

Frequently Asked Questions about effect-uai

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build AI agent loops with streaming and tool execution?

AI agent loops with streaming and tool execution are built using Effect-uai primitives like loop and streamTurn, which handle wire formats and provider integration while you retain ownership of state and flow.

Can I use a provider-agnostic design for multi-provider language model scenarios?

Provider-agnostic design is supported through pluggable language model layers, enabling integration with OpenAI, Anthropic, and Gemini for multi-provider scenarios without enforcing a fixed runtime.

How do I get started creating a minimal streaming turn loop?

To create a minimal streaming turn loop, install the core package alongside a provider package, then scaffold the loop using streamTurn to observe streaming turns and tool outputs.

Does this approach enforce a fixed runtime for managing agent state?

No fixed runtime is enforced; the system handles wire formats, streaming events, and provider integration while giving you full ownership of the agent state and execution flow.

How are structured outputs and tool integrations handled in agent loops?

Structured outputs and tool integrations are handled within agent loops using typed schemas, ensuring typed events and structured outputs flow correctly through the streaming turn.

What is the best way to compose end-to-end agent workflows?

End-to-end agent workflows are composed using reusable recipe catalog patterns, allowing you to combine loop primitives and provider layers into complete agent workflows.