feat: add direct endpoint overrides for auxiliary and delegation

Add base_url/api_key overrides for auxiliary tasks and delegation so users can
route those flows straight to a custom OpenAI-compatible endpoint without
having to rely on provider=main or named custom providers.

Also clear gateway session env vars in test isolation so the full suite stays
deterministic when run from a messaging-backed agent session.

tests/tools/test_delegate.py

@@ -10,6 +10,7 @@ Run with:  python -m pytest tests/test_delegate.py -v
 """
 
 import json
+import os
 import sys
 import unittest
 from unittest.mock import MagicMock, patch
@@ -462,6 +463,43 @@ class TestDelegationCredentialResolution(unittest.TestCase):
         self.assertEqual(creds["api_mode"], "chat_completions")
         mock_resolve.assert_called_once_with(requested="openrouter")
 
+    def test_direct_endpoint_uses_configured_base_url_and_api_key(self):
+        parent = _make_mock_parent(depth=0)
+        cfg = {
+            "model": "qwen2.5-coder",
+            "provider": "openrouter",
+            "base_url": "http://localhost:1234/v1",
+            "api_key": "local-key",
+        }
+        creds = _resolve_delegation_credentials(cfg, parent)
+        self.assertEqual(creds["model"], "qwen2.5-coder")
+        self.assertEqual(creds["provider"], "custom")
+        self.assertEqual(creds["base_url"], "http://localhost:1234/v1")
+        self.assertEqual(creds["api_key"], "local-key")
+        self.assertEqual(creds["api_mode"], "chat_completions")
+
+    def test_direct_endpoint_falls_back_to_openai_api_key_env(self):
+        parent = _make_mock_parent(depth=0)
+        cfg = {
+            "model": "qwen2.5-coder",
+            "base_url": "http://localhost:1234/v1",
+        }
+        with patch.dict(os.environ, {"OPENAI_API_KEY": "env-openai-key"}, clear=False):
+            creds = _resolve_delegation_credentials(cfg, parent)
+        self.assertEqual(creds["api_key"], "env-openai-key")
+        self.assertEqual(creds["provider"], "custom")
+
+    def test_direct_endpoint_does_not_fall_back_to_openrouter_api_key_env(self):
+        parent = _make_mock_parent(depth=0)
+        cfg = {
+            "model": "qwen2.5-coder",
+            "base_url": "http://localhost:1234/v1",
+        }
+        with patch.dict(os.environ, {"OPENROUTER_API_KEY": "env-openrouter-key"}, clear=False):
+            with self.assertRaises(ValueError) as ctx:
+                _resolve_delegation_credentials(cfg, parent)
+        self.assertIn("OPENAI_API_KEY", str(ctx.exception))
+
     @patch("hermes_cli.runtime_provider.resolve_runtime_provider")
     def test_nous_provider_resolves_nous_credentials(self, mock_resolve):
         """Nous provider resolves Nous Portal base_url and api_key."""
@@ -589,6 +627,40 @@ class TestDelegationProviderIntegration(unittest.TestCase):
             self.assertNotEqual(kwargs["base_url"], parent.base_url)
             self.assertNotEqual(kwargs["api_key"], parent.api_key)
 
+    @patch("tools.delegate_tool._load_config")
+    @patch("tools.delegate_tool._resolve_delegation_credentials")
+    def test_direct_endpoint_credentials_reach_child_agent(self, mock_creds, mock_cfg):
+        mock_cfg.return_value = {
+            "max_iterations": 45,
+            "model": "qwen2.5-coder",
+            "base_url": "http://localhost:1234/v1",
+            "api_key": "local-key",
+        }
+        mock_creds.return_value = {
+            "model": "qwen2.5-coder",
+            "provider": "custom",
+            "base_url": "http://localhost:1234/v1",
+            "api_key": "local-key",
+            "api_mode": "chat_completions",
+        }
+        parent = _make_mock_parent(depth=0)
+
+        with patch("run_agent.AIAgent") as MockAgent:
+            mock_child = MagicMock()
+            mock_child.run_conversation.return_value = {
+                "final_response": "done", "completed": True, "api_calls": 1
+            }
+            MockAgent.return_value = mock_child
+
+            delegate_task(goal="Direct endpoint test", parent_agent=parent)
+
+            _, kwargs = MockAgent.call_args
+            self.assertEqual(kwargs["model"], "qwen2.5-coder")
+            self.assertEqual(kwargs["provider"], "custom")
+            self.assertEqual(kwargs["base_url"], "http://localhost:1234/v1")
+            self.assertEqual(kwargs["api_key"], "local-key")
+            self.assertEqual(kwargs["api_mode"], "chat_completions")
+
     @patch("tools.delegate_tool._load_config")
     @patch("tools.delegate_tool._resolve_delegation_credentials")
     def test_empty_config_inherits_parent(self, mock_creds, mock_cfg):