Fix delegation config precedence

tests/tools/test_delegate.py

@@ -13,6 +13,7 @@ import json
 import os
 import threading
 import time
+import types
 import unittest
 from unittest.mock import MagicMock, patch
 
@@ -21,6 +22,7 @@ from tools.delegate_tool import (
     DELEGATE_TASK_SCHEMA,
     DelegateEvent,
     _get_max_concurrent_children,
+    _load_config,
     _LEGACY_EVENT_MAP,
     MAX_DEPTH,
     check_delegate_requirements,
@@ -2390,6 +2392,43 @@ class TestDelegateEventEnum(unittest.TestCase):
 class TestConcurrencyDefaults(unittest.TestCase):
     """Tests for the concurrency default and no hard ceiling."""
 
+    def test_load_config_prefers_active_persistent_config_over_cli_defaults(self):
+        stale_cli = types.ModuleType("cli")
+        stale_cli.CLI_CONFIG = {
+            "delegation": {
+                "max_iterations": 45,
+                "model": "",
+                "provider": "",
+                "base_url": "",
+                "api_key": "",
+            }
+        }
+        active_config = {
+            "delegation": {
+                "max_iterations": 50,
+                "max_concurrent_children": 50,
+                "max_spawn_depth": 10,
+            }
+        }
+
+        with patch.dict("sys.modules", {"cli": stale_cli}):
+            with patch("hermes_cli.config.load_config", return_value=active_config):
+                self.assertEqual(_load_config()["max_concurrent_children"], 50)
+                self.assertEqual(_get_max_concurrent_children(), 50)
+
+    def test_load_config_falls_back_to_cli_config_when_persistent_load_fails(self):
+        fallback_cli = types.ModuleType("cli")
+        fallback_cli.CLI_CONFIG = {
+            "delegation": {
+                "max_iterations": 45,
+                "max_concurrent_children": 8,
+            }
+        }
+
+        with patch.dict("sys.modules", {"cli": fallback_cli}):
+            with patch("hermes_cli.config.load_config", side_effect=RuntimeError("boom")):
+                self.assertEqual(_load_config()["max_concurrent_children"], 8)
+
     @patch("tools.delegate_tool._load_config", return_value={})
     def test_default_is_three(self, mock_cfg):
         # Clear env var if set

tools/delegate_tool.py

@@ -3098,26 +3098,28 @@ def _resolve_delegation_credentials(cfg: dict, parent_agent) -> dict:
 
 
 def _load_config() -> dict:
-    """Load delegation config from CLI_CONFIG or persistent config.
+    """Load delegation config from the active Hermes config.
 
-    Checks the runtime config (cli.py CLI_CONFIG) first, then falls back
-    to the persistent config (hermes_cli/config.py load_config()) so that
-    ``delegation.model`` / ``delegation.provider`` are picked up regardless
-    of the entry point (CLI, gateway, cron).
+    Prefer the shared persistent loader because it follows the active
+    HERMES_HOME/profile. ``cli.CLI_CONFIG`` is a legacy fallback for entry
+    points that cannot import the shared loader; importing it first can return
+    an old default ``delegation`` block and hide user-set keys such as
+    ``max_concurrent_children``.
     """
-    try:
-        from cli import CLI_CONFIG
-
-        cfg = CLI_CONFIG.get("delegation") or {}
-        if cfg:
-            return cfg
-    except Exception:
-        pass
     try:
         from hermes_cli.config import load_config
 
         full = load_config()
-        return full.get("delegation") or {}
+        cfg = full.get("delegation") or {}
+        if isinstance(cfg, dict):
+            return cfg
+    except Exception:
+        pass
+    try:
+        from cli import CLI_CONFIG
+
+        cfg = CLI_CONFIG.get("delegation") or {}
+        return cfg if isinstance(cfg, dict) else {}
     except Exception:
         return {}