feat(dx): Setup Wizard + doctor command — closes #2030 (#2034)

backend/tests/conftest.py

@@ -10,6 +10,7 @@ from unittest.mock import MagicMock
 
 # Make 'app' and 'deerflow' importable from any working directory
 sys.path.insert(0, str(Path(__file__).parent.parent))
+sys.path.insert(0, str(Path(__file__).resolve().parents[2] / "scripts"))
 
 # Break the circular import chain that exists in production code:
 #   deerflow.subagents.__init__

backend/tests/test_doctor.py

@@ -0,0 +1,342 @@
+"""Unit tests for scripts/doctor.py.
+
+Run from repo root:
+    cd backend && uv run pytest tests/test_doctor.py -v
+"""
+
+from __future__ import annotations
+
+import sys
+
+import doctor
+
+# ---------------------------------------------------------------------------
+# check_python
+# ---------------------------------------------------------------------------
+
+
+class TestCheckPython:
+    def test_current_python_passes(self):
+        result = doctor.check_python()
+        assert sys.version_info >= (3, 12)
+        assert result.status == "ok"
+
+
+# ---------------------------------------------------------------------------
+# check_config_exists
+# ---------------------------------------------------------------------------
+
+
+class TestCheckConfigExists:
+    def test_missing_config(self, tmp_path):
+        result = doctor.check_config_exists(tmp_path / "config.yaml")
+        assert result.status == "fail"
+        assert result.fix is not None
+
+    def test_present_config(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\n")
+        result = doctor.check_config_exists(cfg)
+        assert result.status == "ok"
+
+
+# ---------------------------------------------------------------------------
+# check_config_version
+# ---------------------------------------------------------------------------
+
+
+class TestCheckConfigVersion:
+    def test_up_to_date(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\n")
+        example = tmp_path / "config.example.yaml"
+        example.write_text("config_version: 5\n")
+        result = doctor.check_config_version(cfg, tmp_path)
+        assert result.status == "ok"
+
+    def test_outdated(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 3\n")
+        example = tmp_path / "config.example.yaml"
+        example.write_text("config_version: 5\n")
+        result = doctor.check_config_version(cfg, tmp_path)
+        assert result.status == "warn"
+        assert result.fix is not None
+
+    def test_missing_config_skipped(self, tmp_path):
+        result = doctor.check_config_version(tmp_path / "config.yaml", tmp_path)
+        assert result.status == "skip"
+
+
+# ---------------------------------------------------------------------------
+# check_config_loadable
+# ---------------------------------------------------------------------------
+
+
+class TestCheckConfigLoadable:
+    def test_loadable_config(self, tmp_path, monkeypatch):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\n")
+        monkeypatch.setattr(doctor, "_load_app_config", lambda _path: object())
+        result = doctor.check_config_loadable(cfg)
+        assert result.status == "ok"
+
+    def test_invalid_config(self, tmp_path, monkeypatch):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\n")
+
+        def fail(_path):
+            raise ValueError("bad config")
+
+        monkeypatch.setattr(doctor, "_load_app_config", fail)
+        result = doctor.check_config_loadable(cfg)
+        assert result.status == "fail"
+        assert "bad config" in result.detail
+
+
+# ---------------------------------------------------------------------------
+# check_models_configured
+# ---------------------------------------------------------------------------
+
+
+class TestCheckModelsConfigured:
+    def test_no_models(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\nmodels: []\n")
+        result = doctor.check_models_configured(cfg)
+        assert result.status == "fail"
+
+    def test_one_model(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\nmodels:\n  - name: default\n    use: langchain_openai:ChatOpenAI\n    model: gpt-4o\n    api_key: $OPENAI_API_KEY\n")
+        result = doctor.check_models_configured(cfg)
+        assert result.status == "ok"
+
+    def test_missing_config_skipped(self, tmp_path):
+        result = doctor.check_models_configured(tmp_path / "config.yaml")
+        assert result.status == "skip"
+
+
+# ---------------------------------------------------------------------------
+# check_llm_api_key
+# ---------------------------------------------------------------------------
+
+
+class TestCheckLLMApiKey:
+    def test_key_set(self, tmp_path, monkeypatch):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\nmodels:\n  - name: default\n    use: langchain_openai:ChatOpenAI\n    model: gpt-4o\n    api_key: $OPENAI_API_KEY\n")
+        monkeypatch.setenv("OPENAI_API_KEY", "sk-test")
+        results = doctor.check_llm_api_key(cfg)
+        assert any(r.status == "ok" for r in results)
+        assert all(r.status != "fail" for r in results)
+
+    def test_key_missing(self, tmp_path, monkeypatch):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\nmodels:\n  - name: default\n    use: langchain_openai:ChatOpenAI\n    model: gpt-4o\n    api_key: $OPENAI_API_KEY\n")
+        monkeypatch.delenv("OPENAI_API_KEY", raising=False)
+        results = doctor.check_llm_api_key(cfg)
+        assert any(r.status == "fail" for r in results)
+        failed = [r for r in results if r.status == "fail"]
+        assert all(r.fix is not None for r in failed)
+        assert any("OPENAI_API_KEY" in (r.fix or "") for r in failed)
+
+    def test_missing_config_returns_empty(self, tmp_path):
+        results = doctor.check_llm_api_key(tmp_path / "config.yaml")
+        assert results == []
+
+
+# ---------------------------------------------------------------------------
+# check_llm_auth
+# ---------------------------------------------------------------------------
+
+
+class TestCheckLLMAuth:
+    def test_codex_auth_file_missing_fails(self, tmp_path, monkeypatch):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\nmodels:\n  - name: codex\n    use: deerflow.models.openai_codex_provider:CodexChatModel\n    model: gpt-5.4\n")
+        monkeypatch.setenv("CODEX_AUTH_PATH", str(tmp_path / "missing-auth.json"))
+        results = doctor.check_llm_auth(cfg)
+        assert any(result.status == "fail" and "Codex CLI auth available" in result.label for result in results)
+
+    def test_claude_oauth_env_passes(self, tmp_path, monkeypatch):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\nmodels:\n  - name: claude\n    use: deerflow.models.claude_provider:ClaudeChatModel\n    model: claude-sonnet-4-6\n")
+        monkeypatch.setenv("CLAUDE_CODE_OAUTH_TOKEN", "token")
+        results = doctor.check_llm_auth(cfg)
+        assert any(result.status == "ok" and "Claude auth available" in result.label for result in results)
+
+
+# ---------------------------------------------------------------------------
+# check_web_search
+# ---------------------------------------------------------------------------
+
+
+class TestCheckWebSearch:
+    def test_ddg_always_ok(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text(
+            "config_version: 5\nmodels:\n  - name: default\n    use: langchain_openai:ChatOpenAI\n    model: gpt-4o\n    api_key: $OPENAI_API_KEY\ntools:\n  - name: web_search\n    use: deerflow.community.ddg_search.tools:web_search_tool\n"
+        )
+        result = doctor.check_web_search(cfg)
+        assert result.status == "ok"
+        assert "DuckDuckGo" in result.detail
+
+    def test_tavily_with_key_ok(self, tmp_path, monkeypatch):
+        monkeypatch.setenv("TAVILY_API_KEY", "tvly-test")
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\ntools:\n  - name: web_search\n    use: deerflow.community.tavily.tools:web_search_tool\n")
+        result = doctor.check_web_search(cfg)
+        assert result.status == "ok"
+
+    def test_tavily_without_key_warns(self, tmp_path, monkeypatch):
+        monkeypatch.delenv("TAVILY_API_KEY", raising=False)
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\ntools:\n  - name: web_search\n    use: deerflow.community.tavily.tools:web_search_tool\n")
+        result = doctor.check_web_search(cfg)
+        assert result.status == "warn"
+        assert result.fix is not None
+        assert "make setup" in result.fix
+
+    def test_no_search_tool_warns(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\ntools: []\n")
+        result = doctor.check_web_search(cfg)
+        assert result.status == "warn"
+        assert result.fix is not None
+        assert "make setup" in result.fix
+
+    def test_missing_config_skipped(self, tmp_path):
+        result = doctor.check_web_search(tmp_path / "config.yaml")
+        assert result.status == "skip"
+
+    def test_invalid_provider_use_fails(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\ntools:\n  - name: web_search\n    use: deerflow.community.not_real.tools:web_search_tool\n")
+        result = doctor.check_web_search(cfg)
+        assert result.status == "fail"
+
+
+# ---------------------------------------------------------------------------
+# check_web_fetch
+# ---------------------------------------------------------------------------
+
+
+class TestCheckWebFetch:
+    def test_jina_always_ok(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\ntools:\n  - name: web_fetch\n    use: deerflow.community.jina_ai.tools:web_fetch_tool\n")
+        result = doctor.check_web_fetch(cfg)
+        assert result.status == "ok"
+        assert "Jina AI" in result.detail
+
+    def test_firecrawl_without_key_warns(self, tmp_path, monkeypatch):
+        monkeypatch.delenv("FIRECRAWL_API_KEY", raising=False)
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\ntools:\n  - name: web_fetch\n    use: deerflow.community.firecrawl.tools:web_fetch_tool\n")
+        result = doctor.check_web_fetch(cfg)
+        assert result.status == "warn"
+        assert "FIRECRAWL_API_KEY" in (result.fix or "")
+
+    def test_no_fetch_tool_warns(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\ntools: []\n")
+        result = doctor.check_web_fetch(cfg)
+        assert result.status == "warn"
+        assert result.fix is not None
+
+    def test_invalid_provider_use_fails(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\ntools:\n  - name: web_fetch\n    use: deerflow.community.not_real.tools:web_fetch_tool\n")
+        result = doctor.check_web_fetch(cfg)
+        assert result.status == "fail"
+
+
+# ---------------------------------------------------------------------------
+# check_env_file
+# ---------------------------------------------------------------------------
+
+
+class TestCheckEnvFile:
+    def test_missing(self, tmp_path):
+        result = doctor.check_env_file(tmp_path)
+        assert result.status == "warn"
+
+    def test_present(self, tmp_path):
+        (tmp_path / ".env").write_text("KEY=val\n")
+        result = doctor.check_env_file(tmp_path)
+        assert result.status == "ok"
+
+
+# ---------------------------------------------------------------------------
+# check_frontend_env
+# ---------------------------------------------------------------------------
+
+
+class TestCheckFrontendEnv:
+    def test_missing(self, tmp_path):
+        result = doctor.check_frontend_env(tmp_path)
+        assert result.status == "warn"
+
+    def test_present(self, tmp_path):
+        frontend_dir = tmp_path / "frontend"
+        frontend_dir.mkdir()
+        (frontend_dir / ".env").write_text("KEY=val\n")
+        result = doctor.check_frontend_env(tmp_path)
+        assert result.status == "ok"
+
+
+# ---------------------------------------------------------------------------
+# check_sandbox
+# ---------------------------------------------------------------------------
+
+
+class TestCheckSandbox:
+    def test_missing_sandbox_fails(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\n")
+        results = doctor.check_sandbox(cfg)
+        assert results[0].status == "fail"
+
+    def test_local_sandbox_with_disabled_host_bash_warns(self, tmp_path):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\nsandbox:\n  use: deerflow.sandbox.local:LocalSandboxProvider\n  allow_host_bash: false\ntools:\n  - name: bash\n    use: deerflow.sandbox.tools:bash_tool\n")
+        results = doctor.check_sandbox(cfg)
+        assert any(result.status == "warn" for result in results)
+
+    def test_container_sandbox_without_runtime_warns(self, tmp_path, monkeypatch):
+        cfg = tmp_path / "config.yaml"
+        cfg.write_text("config_version: 5\nsandbox:\n  use: deerflow.community.aio_sandbox:AioSandboxProvider\ntools: []\n")
+        monkeypatch.setattr(doctor.shutil, "which", lambda _name: None)
+        results = doctor.check_sandbox(cfg)
+        assert any(result.label == "container runtime available" and result.status == "warn" for result in results)
+
+
+# ---------------------------------------------------------------------------
+# main() exit code
+# ---------------------------------------------------------------------------
+
+
+class TestMainExitCode:
+    def test_returns_int(self, tmp_path, monkeypatch, capsys):
+        """main() should return 0 or 1 without raising."""
+        repo_root = tmp_path / "repo"
+        scripts_dir = repo_root / "scripts"
+        scripts_dir.mkdir(parents=True)
+        fake_doctor = scripts_dir / "doctor.py"
+        fake_doctor.write_text("# test-only shim for __file__ resolution\n")
+
+        monkeypatch.chdir(repo_root)
+        monkeypatch.setattr(doctor, "__file__", str(fake_doctor))
+        monkeypatch.delenv("OPENAI_API_KEY", raising=False)
+        monkeypatch.delenv("TAVILY_API_KEY", raising=False)
+
+        exit_code = doctor.main()
+
+        captured = capsys.readouterr()
+        output = captured.out + captured.err
+
+        assert exit_code in (0, 1)
+        assert output
+        assert "config.yaml" in output
+        assert ".env" in output

backend/tests/test_firecrawl_tools.py

@@ -0,0 +1,66 @@
+"""Unit tests for the Firecrawl community tools."""
+
+import json
+from unittest.mock import MagicMock, patch
+
+
+class TestWebSearchTool:
+    @patch("deerflow.community.firecrawl.tools.FirecrawlApp")
+    @patch("deerflow.community.firecrawl.tools.get_app_config")
+    def test_search_uses_web_search_config(self, mock_get_app_config, mock_firecrawl_cls):
+        search_config = MagicMock()
+        search_config.model_extra = {"api_key": "firecrawl-search-key", "max_results": 7}
+        mock_get_app_config.return_value.get_tool_config.return_value = search_config
+
+        mock_result = MagicMock()
+        mock_result.web = [
+            MagicMock(title="Result", url="https://example.com", description="Snippet"),
+        ]
+        mock_firecrawl_cls.return_value.search.return_value = mock_result
+
+        from deerflow.community.firecrawl.tools import web_search_tool
+
+        result = web_search_tool.invoke({"query": "test query"})
+
+        assert json.loads(result) == [
+            {
+                "title": "Result",
+                "url": "https://example.com",
+                "snippet": "Snippet",
+            }
+        ]
+        mock_get_app_config.return_value.get_tool_config.assert_called_with("web_search")
+        mock_firecrawl_cls.assert_called_once_with(api_key="firecrawl-search-key")
+        mock_firecrawl_cls.return_value.search.assert_called_once_with("test query", limit=7)
+
+
+class TestWebFetchTool:
+    @patch("deerflow.community.firecrawl.tools.FirecrawlApp")
+    @patch("deerflow.community.firecrawl.tools.get_app_config")
+    def test_fetch_uses_web_fetch_config(self, mock_get_app_config, mock_firecrawl_cls):
+        fetch_config = MagicMock()
+        fetch_config.model_extra = {"api_key": "firecrawl-fetch-key"}
+
+        def get_tool_config(name):
+            if name == "web_fetch":
+                return fetch_config
+            return None
+
+        mock_get_app_config.return_value.get_tool_config.side_effect = get_tool_config
+
+        mock_scrape_result = MagicMock()
+        mock_scrape_result.markdown = "Fetched markdown"
+        mock_scrape_result.metadata = MagicMock(title="Fetched Page")
+        mock_firecrawl_cls.return_value.scrape.return_value = mock_scrape_result
+
+        from deerflow.community.firecrawl.tools import web_fetch_tool
+
+        result = web_fetch_tool.invoke({"url": "https://example.com"})
+
+        assert result == "# Fetched Page\n\nFetched markdown"
+        mock_get_app_config.return_value.get_tool_config.assert_any_call("web_fetch")
+        mock_firecrawl_cls.assert_called_once_with(api_key="firecrawl-fetch-key")
+        mock_firecrawl_cls.return_value.scrape.assert_called_once_with(
+            "https://example.com",
+            formats=["markdown"],
+        )

backend/tests/test_setup_wizard.py

@@ -0,0 +1,431 @@
+"""Unit tests for the Setup Wizard (scripts/wizard/).
+
+Run from repo root:
+    cd backend && uv run pytest tests/test_setup_wizard.py -v
+"""
+
+from __future__ import annotations
+
+import yaml
+from wizard.providers import LLM_PROVIDERS, SEARCH_PROVIDERS, WEB_FETCH_PROVIDERS
+from wizard.steps import search as search_step
+from wizard.writer import (
+    build_minimal_config,
+    read_env_file,
+    write_config_yaml,
+    write_env_file,
+)
+
+
+class TestProviders:
+    def test_llm_providers_not_empty(self):
+        assert len(LLM_PROVIDERS) >= 8
+
+    def test_llm_providers_have_required_fields(self):
+        for p in LLM_PROVIDERS:
+            assert p.name
+            assert p.display_name
+            assert p.use
+            assert ":" in p.use, f"Provider '{p.name}' use path must contain ':'"
+            assert p.models
+            assert p.default_model in p.models
+
+    def test_search_providers_have_required_fields(self):
+        for sp in SEARCH_PROVIDERS:
+            assert sp.name
+            assert sp.display_name
+            assert sp.use
+            assert ":" in sp.use
+
+    def test_search_and_fetch_include_firecrawl(self):
+        assert any(provider.name == "firecrawl" for provider in SEARCH_PROVIDERS)
+        assert any(provider.name == "firecrawl" for provider in WEB_FETCH_PROVIDERS)
+
+    def test_web_fetch_providers_have_required_fields(self):
+        for provider in WEB_FETCH_PROVIDERS:
+            assert provider.name
+            assert provider.display_name
+            assert provider.use
+            assert ":" in provider.use
+            assert provider.tool_name == "web_fetch"
+
+    def test_at_least_one_free_search_provider(self):
+        """At least one search provider needs no API key."""
+        free = [sp for sp in SEARCH_PROVIDERS if sp.env_var is None]
+        assert free, "Expected at least one free (no-key) search provider"
+
+    def test_at_least_one_free_web_fetch_provider(self):
+        free = [provider for provider in WEB_FETCH_PROVIDERS if provider.env_var is None]
+        assert free, "Expected at least one free (no-key) web fetch provider"
+
+
+class TestBuildMinimalConfig:
+    def test_produces_valid_yaml(self):
+        content = build_minimal_config(
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="gpt-4o",
+            display_name="OpenAI / gpt-4o",
+            api_key_field="api_key",
+            env_var="OPENAI_API_KEY",
+        )
+        data = yaml.safe_load(content)
+        assert data is not None
+        assert "models" in data
+        assert len(data["models"]) == 1
+        model = data["models"][0]
+        assert model["name"] == "gpt-4o"
+        assert model["use"] == "langchain_openai:ChatOpenAI"
+        assert model["model"] == "gpt-4o"
+        assert model["api_key"] == "$OPENAI_API_KEY"
+
+    def test_gemini_uses_gemini_api_key_field(self):
+        content = build_minimal_config(
+            provider_use="langchain_google_genai:ChatGoogleGenerativeAI",
+            model_name="gemini-2.0-flash",
+            display_name="Gemini",
+            api_key_field="gemini_api_key",
+            env_var="GEMINI_API_KEY",
+        )
+        data = yaml.safe_load(content)
+        model = data["models"][0]
+        assert "gemini_api_key" in model
+        assert model["gemini_api_key"] == "$GEMINI_API_KEY"
+        assert "api_key" not in model
+
+    def test_search_tool_included(self):
+        content = build_minimal_config(
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="gpt-4o",
+            display_name="OpenAI",
+            api_key_field="api_key",
+            env_var="OPENAI_API_KEY",
+            search_use="deerflow.community.tavily.tools:web_search_tool",
+            search_extra_config={"max_results": 5},
+        )
+        data = yaml.safe_load(content)
+        search_tool = next(t for t in data.get("tools", []) if t["name"] == "web_search")
+        assert search_tool["max_results"] == 5
+
+    def test_openrouter_defaults_are_preserved(self):
+        content = build_minimal_config(
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="google/gemini-2.5-flash-preview",
+            display_name="OpenRouter",
+            api_key_field="api_key",
+            env_var="OPENROUTER_API_KEY",
+            extra_model_config={
+                "base_url": "https://openrouter.ai/api/v1",
+                "request_timeout": 600.0,
+                "max_retries": 2,
+                "max_tokens": 8192,
+                "temperature": 0.7,
+            },
+        )
+        data = yaml.safe_load(content)
+        model = data["models"][0]
+        assert model["base_url"] == "https://openrouter.ai/api/v1"
+        assert model["request_timeout"] == 600.0
+        assert model["max_retries"] == 2
+        assert model["max_tokens"] == 8192
+        assert model["temperature"] == 0.7
+
+    def test_web_fetch_tool_included(self):
+        content = build_minimal_config(
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="gpt-4o",
+            display_name="OpenAI",
+            api_key_field="api_key",
+            env_var="OPENAI_API_KEY",
+            web_fetch_use="deerflow.community.jina_ai.tools:web_fetch_tool",
+            web_fetch_extra_config={"timeout": 10},
+        )
+        data = yaml.safe_load(content)
+        fetch_tool = next(t for t in data.get("tools", []) if t["name"] == "web_fetch")
+        assert fetch_tool["timeout"] == 10
+
+    def test_no_search_tool_when_not_configured(self):
+        content = build_minimal_config(
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="gpt-4o",
+            display_name="OpenAI",
+            api_key_field="api_key",
+            env_var="OPENAI_API_KEY",
+        )
+        data = yaml.safe_load(content)
+        tool_names = [t["name"] for t in data.get("tools", [])]
+        assert "web_search" not in tool_names
+        assert "web_fetch" not in tool_names
+
+    def test_sandbox_included(self):
+        content = build_minimal_config(
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="gpt-4o",
+            display_name="OpenAI",
+            api_key_field="api_key",
+            env_var="OPENAI_API_KEY",
+        )
+        data = yaml.safe_load(content)
+        assert "sandbox" in data
+        assert "use" in data["sandbox"]
+        assert data["sandbox"]["use"] == "deerflow.sandbox.local:LocalSandboxProvider"
+        assert data["sandbox"]["allow_host_bash"] is False
+
+    def test_bash_tool_disabled_by_default(self):
+        content = build_minimal_config(
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="gpt-4o",
+            display_name="OpenAI",
+            api_key_field="api_key",
+            env_var="OPENAI_API_KEY",
+        )
+        data = yaml.safe_load(content)
+        tool_names = [t["name"] for t in data.get("tools", [])]
+        assert "bash" not in tool_names
+
+    def test_can_enable_container_sandbox_and_bash(self):
+        content = build_minimal_config(
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="gpt-4o",
+            display_name="OpenAI",
+            api_key_field="api_key",
+            env_var="OPENAI_API_KEY",
+            sandbox_use="deerflow.community.aio_sandbox:AioSandboxProvider",
+            include_bash_tool=True,
+        )
+        data = yaml.safe_load(content)
+        assert data["sandbox"]["use"] == "deerflow.community.aio_sandbox:AioSandboxProvider"
+        assert "allow_host_bash" not in data["sandbox"]
+        tool_names = [t["name"] for t in data.get("tools", [])]
+        assert "bash" in tool_names
+
+    def test_can_disable_write_tools(self):
+        content = build_minimal_config(
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="gpt-4o",
+            display_name="OpenAI",
+            api_key_field="api_key",
+            env_var="OPENAI_API_KEY",
+            include_write_tools=False,
+        )
+        data = yaml.safe_load(content)
+        tool_names = [t["name"] for t in data.get("tools", [])]
+        assert "write_file" not in tool_names
+        assert "str_replace" not in tool_names
+
+    def test_config_version_present(self):
+        content = build_minimal_config(
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="gpt-4o",
+            display_name="OpenAI",
+            api_key_field="api_key",
+            env_var="OPENAI_API_KEY",
+            config_version=5,
+        )
+        data = yaml.safe_load(content)
+        assert data["config_version"] == 5
+
+    def test_cli_provider_does_not_emit_fake_api_key(self):
+        content = build_minimal_config(
+            provider_use="deerflow.models.openai_codex_provider:CodexChatModel",
+            model_name="gpt-5.4",
+            display_name="Codex CLI",
+            api_key_field="api_key",
+            env_var=None,
+        )
+        data = yaml.safe_load(content)
+        model = data["models"][0]
+        assert "api_key" not in model
+
+
+# ---------------------------------------------------------------------------
+# writer.py — env file helpers
+# ---------------------------------------------------------------------------
+
+
+class TestEnvFileHelpers:
+    def test_write_and_read_new_file(self, tmp_path):
+        env_file = tmp_path / ".env"
+        write_env_file(env_file, {"OPENAI_API_KEY": "sk-test123"})
+        pairs = read_env_file(env_file)
+        assert pairs["OPENAI_API_KEY"] == "sk-test123"
+
+    def test_update_existing_key(self, tmp_path):
+        env_file = tmp_path / ".env"
+        env_file.write_text("OPENAI_API_KEY=old-key\n")
+        write_env_file(env_file, {"OPENAI_API_KEY": "new-key"})
+        pairs = read_env_file(env_file)
+        assert pairs["OPENAI_API_KEY"] == "new-key"
+        # Should not duplicate
+        content = env_file.read_text()
+        assert content.count("OPENAI_API_KEY") == 1
+
+    def test_preserve_existing_keys(self, tmp_path):
+        env_file = tmp_path / ".env"
+        env_file.write_text("TAVILY_API_KEY=tavily-val\n")
+        write_env_file(env_file, {"OPENAI_API_KEY": "sk-new"})
+        pairs = read_env_file(env_file)
+        assert pairs["TAVILY_API_KEY"] == "tavily-val"
+        assert pairs["OPENAI_API_KEY"] == "sk-new"
+
+    def test_preserve_comments(self, tmp_path):
+        env_file = tmp_path / ".env"
+        env_file.write_text("# My .env file\nOPENAI_API_KEY=old\n")
+        write_env_file(env_file, {"OPENAI_API_KEY": "new"})
+        content = env_file.read_text()
+        assert "# My .env file" in content
+
+    def test_read_ignores_comments(self, tmp_path):
+        env_file = tmp_path / ".env"
+        env_file.write_text("# comment\nKEY=value\n")
+        pairs = read_env_file(env_file)
+        assert "# comment" not in pairs
+        assert pairs["KEY"] == "value"
+
+
+# ---------------------------------------------------------------------------
+# writer.py — write_config_yaml
+# ---------------------------------------------------------------------------
+
+
+class TestWriteConfigYaml:
+    def test_generated_config_loadable_by_appconfig(self, tmp_path):
+        """The generated config.yaml must be parseable (basic YAML validity)."""
+
+        config_path = tmp_path / "config.yaml"
+        write_config_yaml(
+            config_path,
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="gpt-4o",
+            display_name="OpenAI / gpt-4o",
+            api_key_field="api_key",
+            env_var="OPENAI_API_KEY",
+        )
+        assert config_path.exists()
+        with open(config_path) as f:
+            data = yaml.safe_load(f)
+        assert isinstance(data, dict)
+        assert "models" in data
+
+    def test_copies_example_defaults_for_unconfigured_sections(self, tmp_path):
+        example_path = tmp_path / "config.example.yaml"
+        example_path.write_text(
+            yaml.safe_dump(
+                {
+                    "config_version": 5,
+                    "log_level": "info",
+                    "token_usage": {"enabled": False},
+                    "tool_groups": [{"name": "web"}, {"name": "file:read"}, {"name": "file:write"}, {"name": "bash"}],
+                    "tools": [
+                        {
+                            "name": "web_search",
+                            "group": "web",
+                            "use": "deerflow.community.ddg_search.tools:web_search_tool",
+                            "max_results": 5,
+                        },
+                        {
+                            "name": "web_fetch",
+                            "group": "web",
+                            "use": "deerflow.community.jina_ai.tools:web_fetch_tool",
+                            "timeout": 10,
+                        },
+                        {
+                            "name": "image_search",
+                            "group": "web",
+                            "use": "deerflow.community.image_search.tools:image_search_tool",
+                            "max_results": 5,
+                        },
+                        {"name": "ls", "group": "file:read", "use": "deerflow.sandbox.tools:ls_tool"},
+                        {"name": "write_file", "group": "file:write", "use": "deerflow.sandbox.tools:write_file_tool"},
+                        {"name": "bash", "group": "bash", "use": "deerflow.sandbox.tools:bash_tool"},
+                    ],
+                    "sandbox": {
+                        "use": "deerflow.sandbox.local:LocalSandboxProvider",
+                        "allow_host_bash": False,
+                    },
+                    "summarization": {"max_tokens": 2048},
+                },
+                sort_keys=False,
+            )
+        )
+
+        config_path = tmp_path / "config.yaml"
+        write_config_yaml(
+            config_path,
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="gpt-4o",
+            display_name="OpenAI / gpt-4o",
+            api_key_field="api_key",
+            env_var="OPENAI_API_KEY",
+        )
+        with open(config_path) as f:
+            data = yaml.safe_load(f)
+
+        assert data["log_level"] == "info"
+        assert data["token_usage"]["enabled"] is False
+        assert data["tool_groups"][0]["name"] == "web"
+        assert data["summarization"]["max_tokens"] == 2048
+        assert any(tool["name"] == "image_search" and tool["max_results"] == 5 for tool in data["tools"])
+
+    def test_config_version_read_from_example(self, tmp_path):
+        """write_config_yaml should read config_version from config.example.yaml if present."""
+
+        example_path = tmp_path / "config.example.yaml"
+        example_path.write_text("config_version: 99\n")
+
+        config_path = tmp_path / "config.yaml"
+        write_config_yaml(
+            config_path,
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="gpt-4o",
+            display_name="OpenAI",
+            api_key_field="api_key",
+            env_var="OPENAI_API_KEY",
+        )
+        with open(config_path) as f:
+            data = yaml.safe_load(f)
+        assert data["config_version"] == 99
+
+    def test_model_base_url_from_extra_config(self, tmp_path):
+        config_path = tmp_path / "config.yaml"
+        write_config_yaml(
+            config_path,
+            provider_use="langchain_openai:ChatOpenAI",
+            model_name="google/gemini-2.5-flash-preview",
+            display_name="OpenRouter",
+            api_key_field="api_key",
+            env_var="OPENROUTER_API_KEY",
+            extra_model_config={"base_url": "https://openrouter.ai/api/v1"},
+        )
+        with open(config_path) as f:
+            data = yaml.safe_load(f)
+        assert data["models"][0]["base_url"] == "https://openrouter.ai/api/v1"
+
+
+class TestSearchStep:
+    def test_reuses_api_key_for_same_provider(self, monkeypatch):
+        monkeypatch.setattr(search_step, "print_header", lambda *_args, **_kwargs: None)
+        monkeypatch.setattr(search_step, "print_success", lambda *_args, **_kwargs: None)
+        monkeypatch.setattr(search_step, "print_info", lambda *_args, **_kwargs: None)
+
+        choices = iter([3, 1])
+        prompts: list[str] = []
+
+        def fake_choice(_prompt, _options, default=0):
+            return next(choices)
+
+        def fake_secret(prompt):
+            prompts.append(prompt)
+            return "shared-api-key"
+
+        monkeypatch.setattr(search_step, "ask_choice", fake_choice)
+        monkeypatch.setattr(search_step, "ask_secret", fake_secret)
+
+        result = search_step.run_search_step()
+
+        assert result.search_provider is not None
+        assert result.fetch_provider is not None
+        assert result.search_provider.name == "exa"
+        assert result.fetch_provider.name == "exa"
+        assert result.search_api_key == "shared-api-key"
+        assert result.fetch_api_key == "shared-api-key"
+        assert prompts == ["EXA_API_KEY"]