feat(cli): skip interactive LLM selection when configured via environment (#873)

Setting the LLM env vars now skips the matching CLI selection step and uses
the value, so OpenAI-compatible endpoints (opencode, LM Studio, etc.) and
unattended runs work without prompting. Unset vars are chosen interactively
as before.

  TRADINGAGENTS_LLM_PROVIDER -> skips provider step (still verifies API key)
  TRADINGAGENTS_LLM_BACKEND_URL -> custom endpoint (else provider default)
  TRADINGAGENTS_DEEP_THINK_LLM / _QUICK_THINK_LLM -> skips model step
  TRADINGAGENTS_OUTPUT_LANGUAGE -> skips language step

Builds on the existing TRADINGAGENTS_* config overrides (which already feed
DEFAULT_CONFIG); this wires the CLI to honor them instead of re-prompting.

tests/test_cli_env_skip.py

@@ -0,0 +1,86 @@
+"""Tests for env-driven CLI behavior (#897, #873).
+
+The config-layer override (TRADINGAGENTS_* -> DEFAULT_CONFIG) is covered by
+test_env_overrides.py. These tests cover the CLI layer: an env-configured
+provider/model/language must skip its interactive prompt and use the value.
+"""
+
+import os
+import unittest
+from unittest import mock
+
+import pytest
+
+
+@pytest.mark.unit
+class TestProviderDefaultUrl(unittest.TestCase):
+    def test_known_providers_resolve(self):
+        from cli.utils import provider_default_url
+        self.assertEqual(provider_default_url("openai"), "https://api.openai.com/v1")
+        self.assertEqual(provider_default_url("DeepSeek"), "https://api.deepseek.com")
+        self.assertIsNone(provider_default_url("google"))  # uses SDK default
+
+    def test_unknown_provider_returns_none(self):
+        from cli.utils import provider_default_url
+        self.assertIsNone(provider_default_url("not-a-provider"))
+
+    def test_ollama_honors_base_url_env(self):
+        from cli.utils import provider_default_url
+        with mock.patch.dict(os.environ, {"OLLAMA_BASE_URL": "http://host:1234/v1"}):
+            self.assertEqual(provider_default_url("ollama"), "http://host:1234/v1")
+
+
+@pytest.mark.unit
+class TestCliSkipsPromptsFromEnv(unittest.TestCase):
+    def test_env_config_skips_llm_prompts(self):
+        import cli.main as m
+
+        env = {
+            "TRADINGAGENTS_LLM_PROVIDER": "openai",
+            "TRADINGAGENTS_DEEP_THINK_LLM": "kimi-k2.5",
+            "TRADINGAGENTS_QUICK_THINK_LLM": "deepseek-v4-pro",
+            "TRADINGAGENTS_LLM_BACKEND_URL": "https://opencode.ai/zen/go/v1",
+            "TRADINGAGENTS_OUTPUT_LANGUAGE": "Japanese",
+        }
+        fake_cfg = dict(m.DEFAULT_CONFIG)
+        fake_cfg.update({
+            "llm_provider": "openai",
+            "backend_url": "https://opencode.ai/zen/go/v1",
+            "quick_think_llm": "deepseek-v4-pro",
+            "deep_think_llm": "kimi-k2.5",
+            "output_language": "Japanese",
+        })
+
+        with mock.patch.dict(os.environ, env, clear=False), \
+             mock.patch.object(m, "DEFAULT_CONFIG", fake_cfg), \
+             mock.patch.object(m, "fetch_announcements", return_value=None), \
+             mock.patch.object(m, "display_announcements"), \
+             mock.patch.object(m, "get_ticker", return_value="AAPL"), \
+             mock.patch.object(m, "get_analysis_date", return_value="2026-05-29"), \
+             mock.patch.object(m, "select_analysts", return_value=[]), \
+             mock.patch.object(m, "select_research_depth", return_value=1), \
+             mock.patch.object(m, "ensure_api_key") as ensure_key, \
+             mock.patch.object(m, "select_llm_provider") as prompt_provider, \
+             mock.patch.object(m, "ask_output_language") as prompt_lang, \
+             mock.patch.object(m, "select_shallow_thinking_agent") as prompt_quick, \
+             mock.patch.object(m, "select_deep_thinking_agent") as prompt_deep:
+            sel = m.get_user_selections()
+
+        # None of the LLM selection prompts should have been shown.
+        prompt_provider.assert_not_called()
+        prompt_lang.assert_not_called()
+        prompt_quick.assert_not_called()
+        prompt_deep.assert_not_called()
+        # API key is still verified for the env-configured provider.
+        ensure_key.assert_called_once()
+
+        # The env values flow into the returned selections.
+        self.assertEqual(sel["llm_provider"], "openai")
+        self.assertEqual(sel["backend_url"], "https://opencode.ai/zen/go/v1")
+        self.assertEqual(sel["shallow_thinker"], "deepseek-v4-pro")
+        self.assertEqual(sel["deep_thinker"], "kimi-k2.5")
+        self.assertEqual(sel["output_language"], "Japanese")
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_ollama_base_url.py

@@ -7,6 +7,23 @@ import importlib
 import pytest
 
 
+@pytest.fixture(scope="module", autouse=True)
+def _resync_reloaded_modules():
+    """Restore module state after this file's importlib.reload() calls.
+
+    Several tests below reload ``cli.utils`` to re-evaluate OLLAMA_BASE_URL.
+    That leaves ``cli.main``'s star-imported names (e.g. get_ticker) bound to
+    the pre-reload module objects, which breaks identity checks in unrelated
+    tests that happen to run afterward. Re-sync once on teardown so the reload
+    doesn't leak across test modules.
+    """
+    yield
+    import cli.utils
+    import cli.main
+    importlib.reload(cli.utils)
+    importlib.reload(cli.main)
+
+
 # ---- openai_client side: _resolve_provider_base_url -----------------------