Automation

automation/computer.py

@@ -0,0 +1,59 @@
+"""
+computer.py
+
+Handles all interaction with the local computer.
+"""
+
+import subprocess
+from pathlib import Path
+
+import mss
+from PIL import Image
+
+
+class Computer:
+    def __init__(self):
+        self.generated_file = Path("generated.py")
+        self.screenshot_file = Path("screenshot.png")
+
+    def take_screenshot(self) -> Path:
+        """
+        Captures the current screen.
+        Returns the path to the screenshot.
+        """
+
+        with mss.mss() as sct:
+            monitor = sct.monitors[1]
+            screenshot = sct.grab(monitor)
+
+            image = Image.frombytes(
+                "RGB",
+                screenshot.size,
+                screenshot.rgb,
+            )
+
+            image.save(self.screenshot_file)
+
+        return self.screenshot_file
+
+    def execute(self, code: str) -> dict:
+        """
+        Saves and executes the Python code generated by the AI.
+        """
+
+        self.generated_file.write_text(
+            code,
+            encoding="utf-8",
+        )
+
+        result = subprocess.run(
+            ["python", str(self.generated_file)],
+            capture_output=True,
+            text=True,
+        )
+
+        return {
+            "stdout": result.stdout,
+            "stderr": result.stderr,
+            "returncode": result.returncode,
+        }

automation/main.py

@@ -0,0 +1,59 @@
+"""
+main.py
+
+Friday's main loop.
+"""
+
+from computer import Computer
+from model import Model
+from prompt import SYSTEM_PROMPT
+
+
+def main():
+    computer = Computer()
+    model = Model()
+
+    goal = input("What would you like Friday to do?\n> ")
+
+    while True:
+        # Take the latest screenshot
+        screenshot = computer.take_screenshot()
+
+        # Build the user prompt
+        user_prompt = f"""
+Current Goal:
+{goal}
+
+This is the latest screenshot of the computer.
+
+Figure out the next step.
+
+If the task is complete, simply output:
+
+TASK_COMPLETE
+
+Otherwise, output ONLY executable Python code.
+"""
+
+        # Ask Gemini what to do next
+        response = model.send(
+            SYSTEM_PROMPT,
+            user_prompt,
+            screenshot,
+        )
+
+        # Stop if the task is finished
+        if response.strip() == "TASK_COMPLETE":
+            print("Task completed.")
+            break
+
+        # Execute the generated Python
+        result = computer.execute(response)
+
+        # Optional: print any errors for debugging
+        if result["stderr"]:
+            print(result["stderr"])
+
+
+if __name__ == "__main__":
+    main()

automation/model.py

@@ -0,0 +1,66 @@
+"""
+model.py
+
+Maintains a continuous conversation with Gemini.
+
+Its only job is to:
+1. Send the goal and latest screenshot to Gemini.
+2. Keep the conversation alive.
+3. Return the Python code Gemini generates.
+"""
+
+import os
+
+from google import genai
+
+
+class Model:
+    def __init__(self):
+        api_key = os.getenv("GEMINI_API_KEY")
+
+        if not api_key:
+            raise ValueError("GEMINI_API_KEY not found.")
+
+        self.client = genai.Client(api_key=api_key)
+        self.model = "gemini-2.5-flash"
+
+        # Stores the conversation history so Gemini remembers
+        # the goal and previous screenshots.
+        self.history = []
+
+    def send(
+        self,
+        system_prompt: str,
+        user_prompt: str,
+        screenshot,
+    ) -> str:
+        """
+        Sends the latest computer state to Gemini and returns
+        the Python code it generates.
+        """
+
+        self.history.append(
+            {
+                "role": "user",
+                "parts": [
+                    {"text": f"{system_prompt}\n\n{user_prompt}"},
+                    screenshot,
+                ],
+            }
+        )
+
+        response = self.client.models.generate_content(
+            model=self.model,
+            contents=self.history,
+        )
+
+        self.history.append(
+            {
+                "role": "model",
+                "parts": [
+                    {"text": response.text},
+                ],
+            }
+        )
+
+        return response.text

automation/prompt.py

@@ -0,0 +1,28 @@
+SYSTEM_PROMPT = """
+You are Friday.
+
+You are controlling a Windows computer.
+
+The user will give you a goal.
+
+You will continuously receive screenshots of the computer.
+
+Your job is to accomplish the goal by writing Python code.
+
+The code you write will be executed immediately.
+
+After execution, you will receive another screenshot.
+
+Based on the new screenshot, continue writing Python until the task is complete.
+
+Guidelines:
+
+- Use Python.
+- Use any installed Python libraries.
+- If another library is required, install it.
+- Think step by step.
+- Only output executable Python code.
+- Do not explain your reasoning.
+- Do not wrap your code in Markdown.
+- Do not output anything except Python.
+"""