tauon 0.1.2

A minimal Python agent framework built on tau-ai.

Tauon

A minimal Python agent framework built on tau. Think of it like flue of tau instead of pi.

Beta software. Tauon is a thin layer over tau-ai, which has not yet published a stable public SDK. Until that SDK is released, internal APIs may change without notice. Not recommended for production use.

Quick start

from tauon import define_agent, define_tool, use_model, use_tool


@define_tool
def get_weather(city: str) -> str:
    """Get the current weather for a city."""
    return f"Sunny and 22°C in {city}."


@define_agent
def WeatherAgent() -> str:
    use_model("gpt-4.1-mini")
    use_tool(get_weather)
    return "You are a weather assistant."

export OPENAI_API_KEY=...
uv run tauon run examples/weather.py --message "What's the weather in Paris?"

Models and environment variables

Tauon resolves models from Tau's built-in provider catalog. API keys are read from environment variables only; Tauon does not load saved credentials from ~/.tau/credentials.json or provider preferences from ~/.tau/providers.json.

use_model() patterns

# Bare model name: resolved against the default provider (openai).
use_model("gpt-4.1-mini")  # needs OPENAI_API_KEY

# provider/model syntax: required when the model ID contains "/" or when you
# want to force a specific provider.
use_model("openai-codex/gpt-5.6-luna")  # needs OPENAI_CODEX_ACCESS_TOKEN
use_model("openrouter/qwen/qwen3-14b")  # needs OPENROUTER_API_KEY
use_model("anthropic/claude-sonnet-4-6")  # needs ANTHROPIC_API_KEY

Provider environment variables

Provider	Env var	Example use_model()
openai	OPENAI_API_KEY	use_model("gpt-4.1-mini")
openai-codex	OPENAI_CODEX_ACCESS_TOKEN	use_model("openai-codex/gpt-5.6-luna")
anthropic	ANTHROPIC_API_KEY	use_model("anthropic/claude-sonnet-4-6")
google	GEMINI_API_KEY	use_model("google/gemini-2.5-pro")
deepseek	DEEPSEEK_API_KEY	use_model("deepseek/deepseek-v4-pro")
xai	XAI_API_KEY	use_model("xai/grok-3")
groq	GROQ_API_KEY	use_model("groq/llama-3.1-8b-instant")
cerebras	CEREBRAS_API_KEY	use_model("cerebras/gpt-oss-120b")
nvidia	NVIDIA_API_KEY	use_model("nvidia/llama-3.3-nemotron-super-49b-v1.5")
openrouter	OPENROUTER_API_KEY	use_model("openrouter/qwen/qwen3-14b")
huggingface	HF_TOKEN	use_model("huggingface/Qwen/Qwen3-235B-A22B")
fireworks	FIREWORKS_API_KEY	use_model("fireworks/accounts/fireworks/models/glm-5p1")
together	TOGETHER_API_KEY	use_model("together/Qwen/Qwen3-235B-A22B-Instruct-2507-tput")
mistral	MISTRAL_API_KEY	use_model("mistral/codestral-latest")
minimax	MINIMAX_API_KEY	use_model("minimax/MiniMax-M3")
minimax-cn	MINIMAX_CN_API_KEY	use_model("minimax-cn/MiniMax-M3")
moonshotai	MOONSHOT_API_KEY	use_model("moonshotai/kimi-k2-0711-preview")
moonshotai-cn	MOONSHOT_API_KEY	use_model("moonshotai-cn/kimi-k2-0711-preview")
kimi-code	KIMI_CODE_API_KEY	use_model("kimi-code/kimi-for-coding")
zai	ZAI_API_KEY	use_model("zai/glm-5.1")
xiaomi	XIAOMI_API_KEY	use_model("xiaomi/mimo-v2-pro")
xiaomi-token-plan-cn	XIAOMI_TOKEN_PLAN_CN_API_KEY	use_model("xiaomi-token-plan-cn/mimo-v2-pro")
xiaomi-token-plan-ams	XIAOMI_TOKEN_PLAN_AMS_API_KEY	use_model("xiaomi-token-plan-ams/mimo-v2-pro")
xiaomi-token-plan-sgp	XIAOMI_TOKEN_PLAN_SGP_API_KEY	use_model("xiaomi-token-plan-sgp/mimo-v2-pro")
vercel-ai-gateway	AI_GATEWAY_API_KEY	use_model("vercel-ai-gateway/alibaba/qwen-3-235b")
opencode	OPENCODE_API_KEY	use_model("opencode/deepseek-v4-pro")
opencode-go	OPENCODE_API_KEY	use_model("opencode-go/kimi-k2.7-code")
github-copilot	COPILOT_GITHUB_TOKEN	use_model("github-copilot/claude-opus-4.5")

Bare model names (e.g. use_model("gpt-4.1-mini")) resolve against the default provider, which is openai. Use the provider/model syntax for every other provider. For OpenRouter the prefix is required because model IDs contain /, e.g. use_model("openrouter/qwen/qwen3-14b").

Model names not yet in Tau's built-in catalog fall back to a plain OpenAI-compatible provider using OPENAI_API_KEY (a warning is logged). This keeps newly-released models usable immediately, but it also means a typo'd model name is sent to OpenAI rather than failing at startup — use the provider/model syntax to force a specific provider.

For a provider not in the catalog, or to override the endpoint explicitly, pass api_key and base_url directly to run_agent() as shown in examples/custom_provider.py.