An open API service indexing awesome lists of open source software.

https://github.com/charlesyuan02/legends-za-teambuilder

An actor-critic agent pipeline made w/ Google ADK to suggest a final Pokemon for my team in Legends Z-A.
https://github.com/charlesyuan02/legends-za-teambuilder

adk-python agents

Last synced: about 2 months ago
JSON representation

An actor-critic agent pipeline made w/ Google ADK to suggest a final Pokemon for my team in Legends Z-A.

Awesome Lists containing this project

README

          

# Legends Z-A Teambuilder

![Cover image](docs/cover-image.png)

LLMs still suck at Pokemon — so I fixed that with a generator-critic pipeline.

When I needed a 6th team member for Pokemon Legends Z-A, ChatGPT suggested Pokemon that weren't in the game and Gemini invented Mega Evolutions that don't exist. After correcting them, I realized the recommendations were just my own bias parroted back to me — a failure mode that hits real AI systems too.

So I built this: a generator-critic agentic pipeline using [Google ADK](https://google.github.io/adk-docs/) and Gemini that actually self-corrects.

**How it works:**
1. A **generator** proposes 3 team candidates.
2. A **critic** (with Google Search access) fact-checks every Pokemon, move, item, and mechanic against real game data.
3. A **refiner** fixes what the critic flags.

Loops until the critic finds zero issues, or 5 iterations — whichever comes first.

## Setup

**Prerequisites:** Python 3.13+, [uv](https://docs.astral.sh/uv/)

1. Clone the repo and install dependencies:

```bash
uv sync
```

2. Copy `.env.example` to `.env` and add your Google API key:

```bash
cp .env.example .env
```

```
GOOGLE_API_KEY=your_key_here
```

3. Run:

```bash
uv run main.py
```