diff --git a/src/neurotron/kernel_log_agent.py b/src/neurotron/kernel_log_agent.py
new file mode 100644
index 0000000..bd4d842
--- /dev/null
+++ b/src/neurotron/kernel_log_agent.py
@@ -0,0 +1,42 @@
+# neurotron/kernel_log_agent.py
+
+from pathlib import Path
+from neurotron.logbus import logbus
+
+class KernelLogAgent:
+    name = "kernel.log"
+
+    def __init__(self, cortex, max_lines=256):
+        self.ctx = cortex
+        self.max_lines = max_lines
+        self.buffer = []  # futuramente vai para dashboard pane
+
+    def observe(self):
+        # leitura "bulk" por enquanto
+        try:
+            with open("/dev/kmsg", "r", encoding="utf-8", errors="ignore") as f:
+                for _ in range(self.max_lines):
+                    line = f.readline()
+                    if not line:
+                        break
+                    line = line.strip()
+                    if not line:
+                        continue
+
+                    # guarda em memória (para TUI)
+                    self.buffer.append(line)
+                    if len(self.buffer) > self.max_lines:
+                        self.buffer = self.buffer[-self.max_lines:]
+
+                    # opcional: envia alguns para o Hippocampus
+                    # self.ctx.memory.remember("kernel.msg", {"line": line})
+        except Exception as e:
+            logbus.debug(f"[kernel.log] erro a ler /dev/kmsg: {e}")
+
+    def think(self):
+        # futuro: detectar padrões, warnings, crashes…
+        pass
+
+    def act(self):
+        # futuro: alertar o utilizador, reiniciar serviços…
+        pass
\ No newline at end of file