Add thread safety and agent health checks

- IRC socket writes protected by threading.Lock in agent.py
- Overseer runs periodic health check (30s interval)
- Detects dead agent processes, cleans up resources, announces in #agents

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

agent/agent.py

@@ -118,6 +118,7 @@ class IRCClient:
         self.nick = nick
         self.sock = None
         self.buf = ""
+        self._lock = threading.Lock()
 
     def connect(self):
         self.sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
@@ -127,7 +128,8 @@ class IRCClient:
         self.send(f"USER {self.nick} 0 * :Fireclaw Agent")
 
     def send(self, msg):
-        self.sock.sendall(f"{msg}\r\n".encode("utf-8"))
+        with self._lock:
+            self.sock.sendall(f"{msg}\r\n".encode("utf-8"))
 
     def join(self, channel):
         self.send(f"JOIN {channel}")

src/overseer.ts

@@ -40,6 +40,7 @@ export async function runOverseer(config: OverseerConfig) {
     log(`Cleaned ${cleaned.length} dead agent(s): ${cleaned.join(", ")}`);
   }
 
+  let knownAgents = new Set(listAgents().map((a) => a.name));
   const bot = new IRC.Client();
 
   bot.connect({
@@ -76,6 +77,7 @@ export async function runOverseer(config: OverseerConfig) {
           const name = parts[2];
           bot.say(event.target, `Invoking agent "${name ?? template}" from template "${template}"...`);
           const info = await startAgent(template, { name });
+          knownAgents.add(info.name);
           bot.say(
             event.target,
             `Agent "${info.name}" started: ${info.nick} [${info.model}] (${info.ip})`
@@ -90,6 +92,7 @@ export async function runOverseer(config: OverseerConfig) {
             return;
           }
           await stopAgent(name);
+          knownAgents.delete(name);
           bot.say(event.target, `Agent "${name}" destroyed.`);
           break;
         }
@@ -184,5 +187,23 @@ export async function runOverseer(config: OverseerConfig) {
   process.on("SIGINT", shutdown);
   process.on("SIGTERM", shutdown);
 
+  // Periodic health check — detect and clean up dead agents
+  const HEALTH_CHECK_INTERVAL = 30_000;
+  const healthCheck = () => {
+    const agents = listAgents(); // listAgents already cleans dead entries
+    // Check if any agents were cleaned (listAgents logs and removes dead ones)
+    // We just need to report the ones that disappeared
+    const currentNames = new Set(agents.map((a) => a.name));
+    for (const name of knownAgents) {
+      if (!currentNames.has(name)) {
+        log(`Agent "${name}" died, cleaned up.`);
+        bot.say("#agents", `Agent "${name}" has died and been cleaned up.`);
+      }
+    }
+    knownAgents = currentNames;
+  };
+
+  setInterval(healthCheck, HEALTH_CHECK_INTERVAL);
+
   log("Overseer started. Waiting for commands...");
 }