fix(agent): try fallback providers at init when primary credential pool is exhausted (#17929)

When a provider's credential pool has a single entry in 429-cooldown,
resolve_provider_client returns None and AIAgent.__init__ raises a
misleading RuntimeError suggesting the API key is missing — even when
valid fallback_providers are configured.

This patch makes __init__ iterate the fallback chain before raising,
mirroring the existing in-flight fallback logic in the request loop.
If a fallback resolves, the agent initializes against it and sets
_fallback_activated=True so _restore_primary_runtime can pick the
primary back up after cooldown.

Closes #17929

run_agent.py

@@ -1473,17 +1473,49 @@ class AIAgent:
                                 _env_hint = _pcfg.api_key_env_vars[0]
                         except Exception:
                             pass
+                        # --- Init-time fallback (#17929) ---
+                        _fb_entries = []
+                        if isinstance(fallback_model, list):
+                            _fb_entries = [
+                                f for f in fallback_model
+                                if isinstance(f, dict) and f.get("provider") and f.get("model")
+                            ]
+                        elif isinstance(fallback_model, dict) and fallback_model.get("provider") and fallback_model.get("model"):
+                            _fb_entries = [fallback_model]
+                        _fb_resolved = False
+                        for _fb in _fb_entries:
+                            _fb_client, _fb_model = resolve_provider_client(
+                                _fb["provider"], model=_fb["model"], raw_codex=True,
+                                explicit_base_url=_fb.get("base_url"),
+                                explicit_api_key=_fb.get("api_key"),
+                            )
+                            if _fb_client is not None:
+                                self.provider = _fb["provider"]
+                                self.model = _fb_model or _fb["model"]
+                                self._fallback_activated = True
+                                client_kwargs = {
+                                    "api_key": _fb_client.api_key,
+                                    "base_url": str(_fb_client.base_url),
+                                }
+                                if _provider_timeout is not None:
+                                    client_kwargs["timeout"] = _provider_timeout
+                                if hasattr(_fb_client, "_default_headers") and _fb_client._default_headers:
+                                    client_kwargs["default_headers"] = dict(_fb_client._default_headers)
+                                _fb_resolved = True
+                                break
+                        if not _fb_resolved:
+                            raise RuntimeError(
+                                f"Provider '{_explicit}' is set in config.yaml but no API key "
+                                f"was found. Set the {_env_hint} environment "
+                                f"variable, or switch to a different provider with `hermes model`."
+                            )
+                    if not getattr(self, "_fallback_activated", False):
+                        # No provider configured — reject with a clear message.
                         raise RuntimeError(
-                            f"Provider '{_explicit}' is set in config.yaml but no API key "
-                            f"was found. Set the {_env_hint} environment "
-                            f"variable, or switch to a different provider with `hermes model`."
+                            "No LLM provider configured. Run `hermes model` to "
+                            "select a provider, or run `hermes setup` for first-time "
+                            "configuration."
                         )
-                    # No provider configured — reject with a clear message.
-                    raise RuntimeError(
-                        "No LLM provider configured. Run `hermes model` to "
-                        "select a provider, or run `hermes setup` for first-time "
-                        "configuration."
-                    )
             
             self._client_kwargs = client_kwargs  # stored for rebuilding after interrupt
 
@@ -1536,7 +1568,7 @@ class AIAgent:
         else:
             self._fallback_chain = []
         self._fallback_index = 0
-        self._fallback_activated = False
+        self._fallback_activated = getattr(self, "_fallback_activated", False)
         # Legacy attribute kept for backward compat (tests, external callers)
         self._fallback_model = self._fallback_chain[0] if self._fallback_chain else None
         if self._fallback_chain and not self.quiet_mode:

tests/run_agent/test_init_fallback_on_exhausted_pool.py

@@ -0,0 +1,69 @@
+"""Regression test for #17929: AIAgent.__init__ should try fallback_model
+when primary provider credentials are exhausted."""
+import pytest
+from unittest.mock import patch, MagicMock
+from run_agent import AIAgent
+
+
+def _make_tool_defs():
+    return [{"type": "function", "function": {"name": "web_search",
+             "description": "search", "parameters": {"type": "object", "properties": {}}}}]
+
+
+def _mock_client(api_key="fb-key-1234567890", base_url="https://fb.example.com/v1"):
+    c = MagicMock()
+    c.api_key = api_key
+    c.base_url = base_url
+    c._default_headers = None
+    return c
+
+
+def test_init_tries_fallback_when_primary_returns_none():
+    """When resolve_provider_client returns None for primary but succeeds for
+    a fallback entry, __init__ should NOT raise RuntimeError."""
+    fb = _mock_client()
+
+    def fake_resolve(provider, model=None, raw_codex=False,
+                     explicit_base_url=None, explicit_api_key=None):
+        if provider == "tencent-token-plan":
+            return fb, "kimi2.5"
+        return None, None  # primary exhausted
+
+    with patch("agent.auxiliary_client.resolve_provider_client", side_effect=fake_resolve), \
+         patch("run_agent.get_tool_definitions", return_value=_make_tool_defs()), \
+         patch("run_agent.check_toolset_requirements", return_value={}), \
+         patch("run_agent.OpenAI", return_value=MagicMock()):
+
+        agent = AIAgent(
+            provider="alibaba-coding-plan",
+            model="qwen3.6-plus",
+            api_key=None,
+            base_url=None,
+            quiet_mode=True,
+            skip_context_files=True,
+            skip_memory=True,
+            fallback_model=[{"provider": "tencent-token-plan", "model": "kimi2.5"}],
+        )
+        assert agent.provider == "tencent-token-plan"
+        assert agent.model == "kimi2.5"
+        assert agent._fallback_activated is True
+
+
+def test_init_raises_when_no_fallback_configured():
+    """When primary returns None and no fallback is set, should raise."""
+    with patch("agent.auxiliary_client.resolve_provider_client", return_value=(None, None)), \
+         patch("run_agent.get_tool_definitions", return_value=_make_tool_defs()), \
+         patch("run_agent.check_toolset_requirements", return_value={}), \
+         patch("run_agent.OpenAI", return_value=MagicMock()):
+
+        with pytest.raises(RuntimeError, match="no API key was found"):
+            AIAgent(
+                provider="alibaba-coding-plan",
+                model="qwen3.6-plus",
+                api_key=None,
+                base_url=None,
+                quiet_mode=True,
+                skip_context_files=True,
+                skip_memory=True,
+                fallback_model=None,
+            )