resolve merge conflict

env/attacker.py

@@ -5,39 +5,32 @@ TARGET = "http://127.0.0.1:7860"
 
 
 def brute_force():
-    print("Starting brute force attack...")
     for i in range(5):
-        response = requests.post(
+        requests.post(
             f"{TARGET}/login",
             data={
                 "username": "admin",
                 "password": "wrong"
             }
         )
-        print("Attempt:", i + 1, response.json())
         time.sleep(0.5)
 
 
 def port_scan():
-    print("Starting port scan...")
     endpoints = ["/admin", "/config", "/backup"]
 
     for ep in endpoints:
-        response = requests.get(f"{TARGET}{ep}")
-        print("Scan:", ep, response.status_code)
+        requests.get(f"{TARGET}{ep}")
 
 
 def credential_stuffing():
-    print("Starting credential stuffing...")
-
     passwords = ["admin", "password", "123456"]
 
     for p in passwords:
-        response = requests.post(
+        requests.post(
             f"{TARGET}/login",
             data={"username": "admin", "password": p}
         )
-        print("Attempt:", p, response.json())
 
 
 def simulate_attack():

env/deception.py

@@ -1,7 +1,5 @@
 def deploy_honeypot():
 
-    print("Honeypot deployed")
-
     return {
         "action": "honeypot",
         "status": "deployed"
@@ -10,8 +8,6 @@ def deploy_honeypot():
 
 def fake_database():
 
-    print("Fake database exposed")
-
     return {
         "action": "fake_db",
         "status": "active"
@@ -20,8 +16,6 @@ def fake_database():
 
 def block_attacker(ip):
 
-    print(f"Blocked attacker: {ip}")
-
     return {
         "action": "block",
         "ip": ip

env/env.py

@@ -6,7 +6,7 @@ SERVER = "http://127.0.0.1:7860"
 
 class DeceptionEnv:
 
-    def __init__(self):
+    def _init_(self):
         self.state = {}
         self.done = False
 
@@ -28,7 +28,6 @@ class DeceptionEnv:
         # Detect brute force
         if action == "detect_attack":
             if failed_logins > 3:
-                print("Brute force detected")
                 reward += 0.2
             else:
                 reward -= 0.1
@@ -37,7 +36,6 @@ class DeceptionEnv:
         if action == "detect_attack":
             for r in requests_log:
                 if isinstance(r, dict) and r.get("type") == "port_scan":
-                    print("Port scan detected")
                     reward += 0.2
                     break
 

inference.py

@@ -6,20 +6,20 @@ from openai import OpenAI
 
 random.seed(42)
 
+# Start server
 from env.fake_server import run_server
 threading.Thread(target=run_server, daemon=True).start()
-
 time.sleep(3)
+
 from env.env import DeceptionEnv
 from env.attacker import simulate_attack
 
-
 API_BASE_URL = os.getenv("API_BASE_URL", "https://router.huggingface.co/v1")
 MODEL_NAME = os.getenv("MODEL_NAME", "Qwen/Qwen2.5-7B-Instruct")
 HF_TOKEN = os.getenv("HF_TOKEN")
 
 if not HF_TOKEN:
-    print("[ERROR] HF_TOKEN not set. Please set your Hugging Face token.")
+    print("[ERROR] HF_TOKEN not set.")
     exit(1)
 
 client = OpenAI(
@@ -28,17 +28,20 @@ client = OpenAI(
 )
 
 env = DeceptionEnv()
-state = env.reset()   # reset only once
+state = env.reset()
 
 print("[START] task=ai-deception env=cyber-security model=AI-agent", flush=True)
 
 rewards = []
 history = []
 
-for step in range(1, 4):
+for step in range(1, 10):
 
     simulate_attack()
 
+    # UPDATE STATE BEFORE DECISION (important fix)
+    state = env.state
+
     summary = {
         "failed_logins": state["failed_logins"],
         "port_scans": state.get("port_scans", 0),
@@ -54,7 +57,7 @@ Previous actions:
 Current summary:
 {summary}
 
-STRICT RULES (must follow exactly):
+STRICT RULES:
 
 1. If no previous action → detect_attack
 2. If last action == detect_attack → deploy_honeypot
@@ -77,7 +80,7 @@ block_ip
 
     action = response.choices[0].message.content.strip()
 
-    # deterministic fallback
+    # fallback logic
     if history:
         last = history[-1]
         if last == "detect_attack":
@@ -95,11 +98,15 @@ block_ip
         f"done={str(done).lower()} error=null",
         flush=True
     )
+
+    if done:
+        break
+
+# FINAL SCORE FIX
 score = min(sum(rewards), 1.0)
 
 print(
-    f"[END] success=true steps=3 score={score:.2f} rewards={','.join(f'{r:.2f}' for r in rewards)}",
+    f"[END] success=true steps={len(rewards)} "
+    f"score={score:.2f} rewards={','.join(f'{r:.2f}' for r in rewards)}",
     flush=True
 )
-while True:
-    time.sleep(60)