Back to packageRELEASE DETAILS

0.3.2

PyPI

agentui-starter-pack

Scaffolding CLI for Optimized Agent Stack

Published19 Feb 2026

Direct dependencies7

Transitive dependenciesNot calculated

README

Version 0.3.2 documentation

Agent UI Starter Pack (A2UI)

High-Fidelity Agent-Driven User Interfaces for Google Cloud.

The Agent UI Starter Pack is a professional distribution for developers building high-fidelity AI applications on Gemini. We provide the architectural "Golden Path" for bridging the gap between conversational intelligence and actionable software.

💡 The Core Mission

❌ The Problem: The "Wall of Text"

Conversational AI today is highly intelligent but low-utility. Users are often stuck behind a "Wall of Text"—trying to parse complex stats, project roadmaps, or financial data out of a chat bubble. This leads to Text Fatigue and limits the agent's ability to act as a real tool.

✅ The Solution: From Chatbot to Cockpit

We move users away from talking to a "box" and into an Agent Cockpit. Instead of just sending text, your agent "manifests" high-fidelity, interactive UI components on the fly using the A2UI Protocol.

Software isn't static anymore; it's synthesized by the Agent based on the user's intent.

🎁 What do you get?

By using this starter pack, you aren't just getting template code—you are getting a Production Framework:

Instant 0-to-1: Skip 3 weeks of setting up Vertex AI Auth, NL-to-JSON parsing, and dynamic React rendering.
The Artifact Registry: A library of 20+ premium React components (StatBars, QuizCards, Timeline) that are native to JSON.
The Bridge Orchestrator: A specialized backend that triages "Conversational Intent" and resolves which UI surfaces to manifest.
Observability-by-Default: A built-in "Ops Console" to inspect the raw NDJSON thought-process of the agent in real-time.

🏗️ Core Pillars

🎭 The Face (Front End) - Primary Focus

Role: The Experience. Adaptive surfaces that change based on what the agent is doing.

CLI: agent-ui-starter-pack
Powered by: React, Vite, A2UI Protocol.
Feature: Dynamic A2UI Renderer and a library of high-fidelity components.

⚙️ The Engine (Agent)

Role: The Brain. Internal reasoning and tool execution.

CLI: agent-starter-pack
Powered by: Python, Vertex AI SDK, ADK.
Feature: Native integration with Agent Engine for managed runtimes.

🚀 Key Features

💎 A2UI Protocol Native

The entire stack is built on the Agent-Driven User Interface (A2UI) protocol. Your agent doesn't just send text; it sends structured JSON that manifests as premium UI components (Timelines, Trophies, Quizzes) in real-time.

🔄 State Synchronization

Seamlessly sync agent reasoning steps with frontend state. Build "Human-in-the-loop" workflows where the user can inspect and refine agent actions before they finalize.

Agentic Stack Architecture

🛠️ Usage (Prescribed Examples)

Scaffolding

Create a new project in seconds using our specialized CLIs:

To create the High-Fidelity Front End:

uvx agent-ui-starter-pack create my-ui-project

To hydrate a Figma wireframe into A2UI:

uvx agent-ui-starter-pack hydrate <figma-url>

Local Development

Start the integrated Vite + API bridge:

make dev

Production Deployment

Deploy the full stack to Google Cloud:

make deploy-prod

📁 Repository Structure

/src/a2ui: Core A2UI rendering logic and base components.
/src/backend: The agent engine logic and API bridge.
/src/components: Premium UI building blocks.
/docs: Detailed integration guides and protocol specifications.

License

MIT

Downloads & popularity

Awaiting observations

No metric observations yet

Download counts and popularity will appear here as the collector receives them.

Declared dependencies 7

Native requirements

7 direct targetsNot calculated transitive count

Package / target	Requirement	Relationship
gitpython Environment { "extras": "", "marker": null, "requires_python": ">=3.10", "evaluation": "NOT_EVALUATED" }	`gitpython>=3.1.0`	runtime
google-adk Environment { "extras": "", "marker": null, "requires_python": ">=3.10", "evaluation": "NOT_EVALUATED" }	`google-adk>=0.1.0`	runtime
google-cloud-aiplatform Environment { "extras": "", "marker": null, "requires_python": ">=3.10", "evaluation": "NOT_EVALUATED" }	`google-cloud-aiplatform>=1.70.0`	runtime
pytest-asyncio Environment { "extras": "", "marker": null, "requires_python": ">=3.10", "evaluation": "NOT_EVALUATED" }	`pytest-asyncio>=0.23.0`	runtime
pytest Environment { "extras": "", "marker": null, "requires_python": ">=3.10", "evaluation": "NOT_EVALUATED" }	`pytest>=8.0.0`	runtime
rich Environment { "extras": "", "marker": null, "requires_python": ">=3.10", "evaluation": "NOT_EVALUATED" }	`rich>=13.0.0`	runtime
typer Environment { "extras": "", "marker": null, "requires_python": ">=3.10", "evaluation": "NOT_EVALUATED" }	`typer>=0.9.0`	runtime