feat: add multi-agent and planning CLI flags to inference and expose environment metadata via /info endpoint

inference.py

@@ -429,6 +429,8 @@ def run_episode(
     llm_every: int,
     max_steps: Optional[int],
     verbose: bool = False,
+    coordinator: bool = False,
+    use_planning: bool = False,
 ) -> dict[str, Any]:
     """Run a single episode and emit hackathon-compliant stdout format."""
     task_name = f"gridmind-task-{task_id}"
@@ -450,7 +452,8 @@ def run_episode(
     obs: dict[str, Any] = {}
 
     try:
-        reset_resp = env_client.reset(task_id=task_id, seed=seed)
+        num_buildings = 3 if coordinator else 1
+        reset_resp = env_client.reset(task_id=task_id, seed=seed, num_buildings=num_buildings)
         if reset_resp is None:
             raise RuntimeError("reset failed")
         obs_list = reset_resp.get("observations", [{}])
@@ -483,7 +486,7 @@ def run_episode(
             # C5: World Modeling - Use /simulate when efficiency is low or faults active
             hvac_eff = obs.get("hvac_efficiency", 1.0)
             active_faults_list = obs.get("active_faults", [])
-            use_simulation = not fast_mode and (hvac_eff < 0.7 or len(active_faults_list) > 0)
+            use_simulation = not fast_mode and (use_planning or hvac_eff < 0.7 or len(active_faults_list) > 0)
 
             sim_result = None
             sim_reward = None
@@ -705,6 +708,16 @@ def main() -> None:
         action="store_true",
         help="Enable automatic task curriculum (Theme 4: Self-Improvement)",
     )
+    parser.add_argument(
+        "--coordinator",
+        action="store_true",
+        help="Multi-building coordinator mode: reset with 3 buildings (Theme 1: Multi-Agent)",
+    )
+    parser.add_argument(
+        "--use-planning",
+        action="store_true",
+        help="Force /simulate world-model call on every step (Theme 3: World Modeling)",
+    )
     args = parser.parse_args()
 
     server_proc = start_environment_server(port=7860)
@@ -751,6 +764,8 @@ def main() -> None:
                     llm_every=args.llm_every,
                     max_steps=args.max_steps,
                     verbose=args.verbose,
+                    coordinator=args.coordinator,
+                    use_planning=args.use_planning,
                 )
                 task_scores.append(float(result["score"]))
                 all_results.append(result)

main.go

@@ -158,6 +158,7 @@ func (s *Server) routes() *http.ServeMux {
 	mux.HandleFunc("/tasks", s.handleTasks)
 	mux.HandleFunc("/metrics", s.handleMetrics)
 	mux.HandleFunc("/ws", s.handleWebSocket)
+	mux.HandleFunc("/info", s.handleInfo)
 	// Reverse proxy for dashboard (runs on port 7861 internally)
 	mux.HandleFunc("/dashboard", s.handleDashboardProxy)
 	mux.HandleFunc("/dashboard/", s.handleDashboardProxy)
@@ -879,3 +880,52 @@ func withCORS(next http.Handler) http.Handler {
 		next.ServeHTTP(w, r)
 	})
 }
+
+// handleInfo returns OpenEnv-standard metadata for automated validators and judges.
+func (s *Server) handleInfo(w http.ResponseWriter, r *http.Request) {
+	if r.Method != http.MethodGet {
+		http.Error(w, "method not allowed", http.StatusMethodNotAllowed)
+		return
+	}
+	info := map[string]interface{}{
+		"name":        "gridmind-rl",
+		"version":     "2.0.0",
+		"description": "Multi-building industrial energy management RL environment with instruction-following, world modeling, fault injection, and curriculum learning.",
+		"multi_agent": true,
+		"themes": []string{
+			"multi-agent",
+			"long-horizon-planning",
+			"world-modeling",
+			"self-improvement",
+		},
+		"observation_space": map[string]interface{}{
+			"type": "dict",
+			"fields": []string{
+				"indoor_temperature", "thermal_storage_level", "current_price",
+				"grid_stress_signal", "carbon_intensity", "hour_of_day", "step",
+				"hvac_efficiency", "process_demand", "cumulative_cost",
+				"batch_queue", "active_faults", "instruction_card",
+			},
+		},
+		"action_space": map[string]interface{}{
+			"type": "dict",
+			"fields": map[string]string{
+				"hvac_power_level":    "float [0.0, 1.0]",
+				"thermal_charge_rate": "float [-1.0, 1.0]",
+				"batch_job_slot":      "int [0, 4]",
+				"load_shed_fraction":  "float [0.0, 0.5]",
+				"building_id":         "int [0, N_buildings-1]",
+			},
+		},
+		"endpoints": []string{
+			"POST /reset", "POST /step", "GET /grade", "GET /tasks",
+			"GET /state", "POST /simulate", "GET /feeder", "POST /coordinate",
+			"GET /health", "GET /info",
+		},
+		"hf_space": "https://lo-kyu-gridmind.hf.space",
+		"github":   "https://github.com/LO-Kyu/gridmind",
+	}
+	w.Header().Set("Content-Type", "application/json")
+	w.Header().Set("Access-Control-Allow-Origin", "*")
+	json.NewEncoder(w).Encode(info)
+}