Skip to content

AI-powered virtual executive team — a single coherent executive persona backed by 8 specialist agents (FastAPI + Next.js).

892 B
Raw Blame
"""LLM provider abstraction.

All Anthropic-shaped LLM calls go through ``get_provider(model)`` rather
than instantiating ``anthropic.AsyncAnthropic`` directly. Today the only
backend is Anthropic itself; the layer exists so a Claude slug can be
routed via OpenRouter (and non-Claude OpenRouter slugs can be selected
per-agent) without rewriting every call site's Anthropic-shaped request.
"""
from openexecutive.providers.provider import LLMProvider
from openexecutive.providers.registry import (
    ANTHROPIC_DIRECT_MODELS,
    OPENROUTER_MODELS,
    allowed_models,
    allowed_models_for,
    get_provider,
    model_options_for,
    model_supports_deep_reasoning,
)

__all__ = [
    "ANTHROPIC_DIRECT_MODELS",
    "LLMProvider",
    "OPENROUTER_MODELS",
    "allowed_models",
    "allowed_models_for",
    "get_provider",
    "model_options_for",
    "model_supports_deep_reasoning",
]