from fastmind import FastMind, Graph, Event
from fastmind.contrib import FastMindAPI

app = FastMind()

@app.agent(name="chat_agent")
async def chat_agent(state: dict, event: Event) -> dict:
    state.setdefault("messages", [])
    state["messages"].append({"role": "user", "content": event.payload.get("text", "")})
    state["messages"].append({"role": "assistant", "content": "Hello!"})
    return state

graph = Graph()
graph.add_node("chat", chat_agent)
graph.set_entry_point("chat")
app.register_graph("main", graph)

async def main():
    api = FastMindAPI(app)
    await api.start()
    await api.push_event("user_001", Event("user.message", {"text": "Hello!"}, "user_001"))
    await api.stop()

import asyncio
asyncio.run(main())

from fastmind import FastMind, Graph, Event, ToolNode, Tool
from fastmind.contrib import FastMindAPI

app = FastMind()

@app.tool(name="get_weather", description="Get weather for a city")
async def get_weather(city: str) -> str:
    return f"{city} is sunny, 20°C"

async def chat_agent(state: dict, event: Event) -> dict:
    state.setdefault("messages", [])
    state["messages"].append({"role": "user", "content": event.payload.get("text", "")})
    # Simulate LLM deciding to call a tool
    if "weather" in event.payload.get("text", "").lower():
        state["tool_calls"] = [
            {"id": "call_1", "function": {"name": "get_weather", "arguments": '{"city": "Beijing"}'}}
        ]
    else:
        state["messages"].append({"role": "assistant", "content": "I can check weather for you!"})
    return state

tool_node = ToolNode(app.get_tools())

def has_tool_calls(state: dict, event: Event) -> str:
    return "tools" if state.get("tool_calls") else None

graph = Graph()
graph.add_node("agent", chat_agent)
graph.add_node("tools", tool_node)
graph.add_conditional_edges("agent", has_tool_calls, {"tools": "tools", None: "__end__"})
graph.add_edge("tools", "agent")
graph.set_entry_point("agent")
app.register_graph("main", graph)

from fastmind import FastMind, Graph, Event, ActionSpace
from fastmind.contrib import FastMindAPI

app = FastMind()

# High-frequency sensor signal (30fps, zero-copy)
@app.signal(name="vision", interval=1/30)
async def npc_vision():
    return {"frame_id": 1, "objects": []}

# VLA fast loop (30Hz, time-driven, bypasses graph)
@app.vla(name="navigation", frequency=30.0)
async def navigation_vla(state, signal_bus):
    vision = signal_bus.read("vision")
    goal = state.get("llm", {}).get("goal", "idle")
    action = [0.5, 0.0, 0.0]  # mock: move forward
    return {"body": action}

# Action executor receives routed action vector
@app.vla_action(name="body", action_space=ActionSpace(3))
async def body_executor(action):
    await game_engine.move(action[0], action[1], action[2])

# LLM slow loop (event-driven)
@app.agent(name="npc_brain")
async def npc_brain(state, event):
    if event.type == "user.message":
        state.setdefault("llm", {})["goal"] = "go_to_castle"
    return state

graph = Graph()
graph.add_node("brain", npc_brain)
graph.set_entry_point("brain")
app.register_graph("main", graph)

api = FastMindAPI(app)

await api.start()
await api.push_event(session_id, event)
async for ev in api.stream_events(session_id): ...

# New VLA/Signal API methods:
frame = api.read_signal(session_id, "vision")      # read signal
api.write_signal(session_id, "gps", data)           # write signal
signals = api.list_signals(session_id)              # list signals
api.pause_vla(session_id)                            # pause VLA loop
api.resume_vla(session_id)                           # resume VLA loop

await api.stop()