feat: single-command Docker startup, fix UI bugs, add Selenium tests

- Add `make up` / `make up DEV=1` for one-command Docker startup with optional hot-reload via docker-compose.dev.yml overlay - Add `timmy up --dev` / `timmy down` CLI commands - Fix cross-platform font resolution in creative assembler (7 test failures) - Fix Ollama host URL not passed to Agno model (container connectivity) - Fix task panel route shadowing by reordering literal routes before parameterized routes in swarm.py - Fix chat input not clearing after send (hx-on::after-request) - Fix chat scroll overflow (CSS min-height: 0 on flex children) - Add Selenium UI smoke tests (17 tests, gated behind SELENIUM_UI=1) - Install fonts-dejavu-core in Dockerfile for container font support - Remove obsolete docker-compose version key - Bump CSS cache-bust to v4 833 unit tests pass, 15 Selenium tests pass (2 skipped). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-25 07:20:56 -05:00
parent 6bae3af268
commit 29292cfb84
16 changed files with 533 additions and 67 deletions
--- a/2
+++ b/2
@@ -15,7 +15,7 @@ FROM python:3.12-slim AS base

 # ── System deps ──────────────────────────────────────────────────────────────
 RUN apt-get update && apt-get install -y --no-install-recommends \
-        gcc curl \
+        gcc curl fonts-dejavu-core \
    && rm -rf /var/lib/apt/lists/*

 WORKDIR /app
--- a/39
+++ b/39
@@ -1,4 +1,5 @@
 .PHONY: install install-bigbrain dev test test-cov test-cov-html watch lint clean help \
+        up down logs \
        docker-build docker-up docker-down docker-agent docker-logs docker-shell \
        cloud-deploy cloud-up cloud-down cloud-logs cloud-status cloud-update

@@ -77,6 +78,33 @@ lint:

 # ── Housekeeping ──────────────────────────────────────────────────────────────

+# ── One-command startup ──────────────────────────────────────────────────────
+#   make up           build + start everything in Docker
+#   make up DEV=1     same, with hot-reload on Python/template/CSS changes
+
+up:
+	mkdir -p data
+ifdef DEV
+	docker compose -f docker-compose.yml -f docker-compose.dev.yml up -d --build
+	@echo ""
+	@echo "  ✓ Timmy Time running in DEV mode at http://localhost:8000"
+	@echo "    Hot-reload active — Python, template, and CSS changes auto-apply"
+	@echo "    Logs: make logs"
+	@echo ""
+else
+	docker compose up -d --build
+	@echo ""
+	@echo "  ✓ Timmy Time running at http://localhost:8000"
+	@echo "    Logs: make logs"
+	@echo ""
+endif
+
+down:
+	docker compose down
+
+logs:
+	docker compose logs -f
+
 # ── Docker ────────────────────────────────────────────────────────────────────

 docker-build:
@@ -150,12 +178,19 @@ clean:
 	rm -rf .pytest_cache htmlcov .coverage coverage.xml

 help:
+	@echo ""
+	@echo "  Quick Start"
+	@echo "  ─────────────────────────────────────────────────"
+	@echo "  make up               build + start everything in Docker"
+	@echo "  make up DEV=1         same, with hot-reload on file changes"
+	@echo "  make down             stop all containers"
+	@echo "  make logs             tail container logs"
 	@echo ""
 	@echo "  Local Development"
 	@echo "  ─────────────────────────────────────────────────"
 	@echo "  make install          create venv + install dev deps"
 	@echo "  make install-bigbrain install with AirLLM (big-model backend)"
-	@echo "  make dev              start dashboard at http://localhost:8000"
+	@echo "  make dev              start dashboard locally (no Docker)"
 	@echo "  make ip               print local IP addresses for phone testing"
 	@echo "  make test             run all tests"
 	@echo "  make test-cov         tests + coverage report (terminal + XML)"
@@ -164,7 +199,7 @@ help:
 	@echo "  make lint             run ruff or flake8"
 	@echo "  make clean            remove build artefacts and caches"
 	@echo ""
-	@echo "  Docker (Dev)"
+	@echo "  Docker (Advanced)"
 	@echo "  ─────────────────────────────────────────────────"
 	@echo "  make docker-build     build the timmy-time:latest image"
 	@echo "  make docker-up        start dashboard container"
--- a/docker-compose.dev.yml
+++ b/docker-compose.dev.yml
@@ -0,0 +1,23 @@
+# ── Timmy Time — Dev-mode overlay ────────────────────────────────────────────
+#
+# Enables hot-reload: Python, template, and CSS changes auto-apply.
+#
+# Usage:
+#   make up DEV=1
+#   # or directly:
+#   docker compose -f docker-compose.yml -f docker-compose.dev.yml up -d --build
+
+services:
+  dashboard:
+    command:
+      - uvicorn
+      - dashboard.app:app
+      - --host=0.0.0.0
+      - --port=8000
+      - --reload
+      - --reload-dir=/app/src
+      - --reload-include=*.html
+      - --reload-include=*.css
+      - --reload-include=*.js
+    environment:
+      DEBUG: "true"
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -15,8 +15,6 @@
 #   make docker-down     stop everything
 #   make docker-logs     tail logs

-version: "3.9"
-
 services:

  # ── Dashboard (coordinator + FastAPI) ──────────────────────────────────────
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -32,6 +32,7 @@ dev = [
    "pytest>=8.0.0",
    "pytest-asyncio>=0.24.0",
    "pytest-cov>=5.0.0",
+    "selenium>=4.20.0",
 ]
 # Big-brain: run 8B / 70B / 405B models locally via layer-by-layer loading.
 # pip install ".[bigbrain]"
--- a/src/creative/assembler.py
+++ b/src/creative/assembler.py
@@ -28,8 +28,26 @@ try:
 except ImportError:
    _MOVIEPY_AVAILABLE = False

-# Resolve a font that actually exists on this system.
-_DEFAULT_FONT = "/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf"
+def _resolve_font() -> str:
+    """Find a usable TrueType font on the current platform."""
+    candidates = [
+        # Linux (Debian/Ubuntu)
+        "/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf",
+        "/usr/share/fonts/TTF/DejaVuSans.ttf",  # Arch
+        "/usr/share/fonts/dejavu-sans-fonts/DejaVuSans.ttf",  # Fedora
+        # macOS
+        "/System/Library/Fonts/Supplemental/Arial.ttf",
+        "/System/Library/Fonts/Helvetica.ttc",
+        "/Library/Fonts/Arial.ttf",
+    ]
+    for path in candidates:
+        if Path(path).exists():
+            return path
+    logger.warning("No system TrueType font found; using Pillow default")
+    return "Helvetica"
+
+
+_DEFAULT_FONT = _resolve_font()


 def _require_moviepy() -> None:
--- a/src/dashboard/routes/swarm.py
+++ b/src/dashboard/routes/swarm.py
@@ -114,6 +114,52 @@ async def post_task_and_auction(description: str = Form(...)):
    }


+@router.get("/tasks/panel", response_class=HTMLResponse)
+async def task_create_panel(request: Request, agent_id: Optional[str] = None):
+    """Task creation panel, optionally pre-selecting an agent."""
+    agents = coordinator.list_swarm_agents()
+    return templates.TemplateResponse(
+        request,
+        "partials/task_assign_panel.html",
+        {"agents": agents, "preselected_agent_id": agent_id},
+    )
+
+
+@router.post("/tasks/direct", response_class=HTMLResponse)
+async def direct_assign_task(
+    request: Request,
+    description: str = Form(...),
+    agent_id: Optional[str] = Form(None),
+):
+    """Create a task: assign directly if agent_id given, else open auction."""
+    timestamp = datetime.now(timezone.utc).strftime("%H:%M:%S")
+
+    if agent_id:
+        agent = registry.get_agent(agent_id)
+        task = coordinator.post_task(description)
+        coordinator.auctions.open_auction(task.id)
+        coordinator.auctions.submit_bid(task.id, agent_id, 1)
+        coordinator.auctions.close_auction(task.id)
+        update_task(task.id, status=TaskStatus.ASSIGNED, assigned_agent=agent_id)
+        registry.update_status(agent_id, "busy")
+        agent_name = agent.name if agent else agent_id
+    else:
+        task = coordinator.post_task(description)
+        winner = await coordinator.run_auction_and_assign(task.id)
+        task = coordinator.get_task(task.id)
+        agent_name = winner.agent_id if winner else "unassigned"
+
+    return templates.TemplateResponse(
+        request,
+        "partials/task_result.html",
+        {
+            "task": task,
+            "agent_name": agent_name,
+            "timestamp": timestamp,
+        },
+    )
+
+
@router.get("/tasks/{task_id}")
 async def get_task(task_id: str):
    """Get details for a specific task."""
@@ -268,47 +314,3 @@ async def message_agent(agent_id: str, request: Request, message: str = Form(...
    )


-@router.get("/tasks/panel", response_class=HTMLResponse)
-async def task_create_panel(request: Request, agent_id: Optional[str] = None):
-    """Task creation panel, optionally pre-selecting an agent."""
-    agents = coordinator.list_swarm_agents()
-    return templates.TemplateResponse(
-        request,
-        "partials/task_assign_panel.html",
-        {"agents": agents, "preselected_agent_id": agent_id},
-    )
-
-
-@router.post("/tasks/direct", response_class=HTMLResponse)
-async def direct_assign_task(
-    request: Request,
-    description: str = Form(...),
-    agent_id: Optional[str] = Form(None),
-):
-    """Create a task: assign directly if agent_id given, else open auction."""
-    timestamp = datetime.now(timezone.utc).strftime("%H:%M:%S")
-
-    if agent_id:
-        agent = registry.get_agent(agent_id)
-        task = coordinator.post_task(description)
-        coordinator.auctions.open_auction(task.id)
-        coordinator.auctions.submit_bid(task.id, agent_id, 1)
-        coordinator.auctions.close_auction(task.id)
-        update_task(task.id, status=TaskStatus.ASSIGNED, assigned_agent=agent_id)
-        registry.update_status(agent_id, "busy")
-        agent_name = agent.name if agent else agent_id
-    else:
-        task = coordinator.post_task(description)
-        winner = await coordinator.run_auction_and_assign(task.id)
-        task = coordinator.get_task(task.id)
-        agent_name = winner.agent_id if winner else "unassigned"
-
-    return templates.TemplateResponse(
-        request,
-        "partials/task_result.html",
-        {
-            "task": task,
-            "agent_name": agent_name,
-            "timestamp": timestamp,
-        },
-    )
--- a/src/dashboard/templates/base.html
+++ b/src/dashboard/templates/base.html
@@ -11,7 +11,7 @@
  <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin />
  <link href="https://fonts.googleapis.com/css2?family=JetBrains+Mono:wght@300;400;500;700&display=swap" rel="stylesheet" />
  <link href="https://cdn.jsdelivr.net/npm/bootstrap@5.3.3/dist/css/bootstrap.min.css" rel="stylesheet" integrity="sha384-QWTKZyjpPEjISv5WaRU9OFeRpok6YctnYmDr5pNlyT2bRjXh0JMhjY6hW+ALEwIH" crossorigin="anonymous" />
-  <link rel="stylesheet" href="/static/style.css?v=3" />
+  <link rel="stylesheet" href="/static/style.css?v=4" />
  {% block extra_styles %}{% endblock %}
  <script src="https://unpkg.com/htmx.org@2.0.3" integrity="sha384-0895/pl2MU10Hqc6jd4RvrthNlDiE9U1tWmX7WRESftEDRosgxNsQG/Ze9YMRzHq" crossorigin="anonymous"></script>
 </head>
--- a/src/dashboard/templates/partials/agent_panel.html
+++ b/src/dashboard/templates/partials/agent_panel.html
@@ -52,7 +52,7 @@
            hx-swap="beforeend"
            hx-indicator="#agent-send-indicator"
            hx-disabled-elt="find button"
-            hx-on::after-settle="this.reset(); scrollAgentLog('{{ agent.id }}')"
+            hx-on::after-request="if(event.detail.successful){this.querySelector('[name=message]').value=''; scrollAgentLog('{{ agent.id }}')}"
            class="d-flex gap-2">
        <input type="text"
               name="message"
--- a/src/dashboard/templates/partials/timmy_panel.html
+++ b/src/dashboard/templates/partials/timmy_panel.html
@@ -18,7 +18,8 @@
    <div class="chat-log flex-grow-1 overflow-auto p-3" id="chat-log"
         hx-get="/agents/timmy/history"
         hx-trigger="load"
-         hx-swap="innerHTML"></div>
+         hx-swap="innerHTML"
+         hx-on::after-settle="scrollChat()"></div>

    <div class="card-footer mc-chat-footer">
      <form hx-post="/agents/timmy/chat"
@@ -27,7 +28,7 @@
            hx-indicator="#send-indicator"
            hx-sync="this:drop"
            hx-disabled-elt="find button"
-            hx-on::after-settle="this.reset(); scrollChat()"
+            hx-on::after-request="if(event.detail.successful){this.querySelector('[name=message]').value=''; scrollChat()}"
            class="d-flex gap-2">
        <input type="text"
               name="message"
--- a/src/timmy/agent.py
+++ b/src/timmy/agent.py
@@ -68,7 +68,7 @@ def create_timmy(
    
    return Agent(
        name="Timmy",
-        model=Ollama(id=settings.ollama_model),
+        model=Ollama(id=settings.ollama_model, host=settings.ollama_url),
        db=SqliteDb(db_file=db_file),
        description=TIMMY_SYSTEM_PROMPT,
        add_history_to_context=True,
--- a/src/timmy/cli.py
+++ b/src/timmy/cli.py
@@ -1,3 +1,4 @@
+import subprocess
 from typing import Optional

 import typer
@@ -54,5 +55,34 @@ def status(
    timmy.print_response(TIMMY_STATUS_PROMPT, stream=False)


+@app.command()
+def up(
+    dev: bool = typer.Option(False, "--dev", help="Enable hot-reload for development"),
+    build: bool = typer.Option(True, "--build/--no-build", help="Rebuild images before starting"),
+):
+    """Start Timmy Time in Docker (dashboard + agents)."""
+    cmd = ["docker", "compose"]
+    if dev:
+        cmd += ["-f", "docker-compose.yml", "-f", "docker-compose.dev.yml"]
+    cmd += ["up", "-d"]
+    if build:
+        cmd.append("--build")
+
+    mode = "dev mode (hot-reload active)" if dev else "production mode"
+    typer.echo(f"Starting Timmy Time in {mode}...")
+    result = subprocess.run(cmd)
+    if result.returncode == 0:
+        typer.echo(f"\n  Timmy Time running at http://localhost:8000  ({mode})\n")
+    else:
+        typer.echo("Failed to start. Is Docker running?", err=True)
+        raise typer.Exit(1)
+
+
+@app.command()
+def down():
+    """Stop all Timmy Time Docker containers."""
+    subprocess.run(["docker", "compose", "down"], check=True)
+
+
 def main():
    app()
--- a/static/style.css
+++ b/static/style.css
@@ -272,7 +272,9 @@ a:hover { color: var(--orange); }
 }
 .mc-chat-panel > .card {
  height: 100%;
-  overflow: clip;
+  min-height: 0;            /* flex child: allow shrink below content */
+  margin-bottom: 0;         /* override Bootstrap .card margin that breaks flex sizing */
+  overflow: hidden;
 }

 /* ── Panel / Card overrides ──────────────────────── */
@@ -396,6 +398,7 @@ a:hover { color: var(--orange); }
 /* ── Chat ────────────────────────────────────────── */
 .chat-log {
  flex: 1;
+  min-height: 0;            /* allow flex child to shrink below content height */
  overflow-y: auto;
  -webkit-overflow-scrolling: touch;
 }
--- a/tests/functional/test_ui_selenium.py
+++ b/tests/functional/test_ui_selenium.py
@@ -0,0 +1,319 @@
+"""Selenium UI tests — green-path smoke tests for the dashboard.
+
+Requires:
+  - Dashboard running at http://localhost:8000 (make up DEV=1)
+  - Chrome installed (headless mode, no display needed)
+  - selenium pip package
+
+Run:
+  SELENIUM_UI=1 pytest tests/functional/test_ui_selenium.py -v --override-ini='pythonpath='
+"""
+
+import os
+import sys
+import time
+
+import pytest
+
+# Skip entire module unless SELENIUM_UI=1 is set
+pytestmark = pytest.mark.skipif(
+    os.environ.get("SELENIUM_UI") != "1",
+    reason="Set SELENIUM_UI=1 to run Selenium UI tests",
+)
+
+DASHBOARD_URL = os.environ.get("DASHBOARD_URL", "http://localhost:8000")
+
+
+# ── Prevent src/websocket from shadowing the websocket-client package ────────
+# Selenium depends on websocket-client which provides `from websocket import
+# WebSocketApp`.  The project's src/websocket/ module would shadow that import.
+# Remove "src" from sys.path for this module since we don't import project code.
+_src_paths = [p for p in sys.path if p.endswith("/src") or p.endswith("\\src")]
+for _p in _src_paths:
+    sys.path.remove(_p)
+
+from selenium import webdriver  # noqa: E402
+from selenium.webdriver.chrome.options import Options  # noqa: E402
+from selenium.webdriver.common.by import By  # noqa: E402
+from selenium.webdriver.common.keys import Keys  # noqa: E402
+from selenium.webdriver.support import expected_conditions as EC  # noqa: E402
+from selenium.webdriver.support.ui import WebDriverWait  # noqa: E402
+
+# Restore paths so other test modules aren't affected
+for _p in _src_paths:
+    if _p not in sys.path:
+        sys.path.append(_p)
+
+
+@pytest.fixture(scope="module")
+def driver():
+    """Headless Chrome WebDriver, shared across tests in this module."""
+    opts = Options()
+    opts.add_argument("--headless=new")
+    opts.add_argument("--no-sandbox")
+    opts.add_argument("--disable-dev-shm-usage")
+    opts.add_argument("--disable-gpu")
+    opts.add_argument("--window-size=1280,900")
+
+    d = webdriver.Chrome(options=opts)
+    d.implicitly_wait(5)
+    yield d
+    d.quit()
+
+
+@pytest.fixture(autouse=True)
+def _check_dashboard():
+    """Skip all tests if the dashboard isn't reachable."""
+    import httpx
+
+    try:
+        r = httpx.get(f"{DASHBOARD_URL}/health", timeout=5)
+        if r.status_code != 200:
+            pytest.skip("Dashboard not healthy")
+    except Exception:
+        pytest.skip("Dashboard not reachable at " + DASHBOARD_URL)
+
+
+def _load_dashboard(driver):
+    """Navigate to dashboard and wait for Timmy panel to load."""
+    driver.get(DASHBOARD_URL)
+    WebDriverWait(driver, 15).until(
+        EC.presence_of_element_located(
+            (By.XPATH, "//*[contains(text(), 'TIMMY INTERFACE')]")
+        )
+    )
+
+
+def _wait_for_sidebar(driver):
+    """Wait for the agent sidebar to finish its HTMX load."""
+    WebDriverWait(driver, 15).until(
+        EC.presence_of_element_located(
+            (By.XPATH, "//*[contains(text(), 'SWARM AGENTS')]")
+        )
+    )
+
+
+def _has_registered_agents(driver):
+    """Check if there are any registered agent cards in the sidebar."""
+    cards = driver.find_elements(By.CSS_SELECTOR, ".mc-agent-card")
+    return len(cards) > 0
+
+
+def _send_chat_and_wait(driver, message):
+    """Send a chat message and wait for the NEW agent response.
+
+    Returns the number of agent messages before and after sending.
+    """
+    existing = len(driver.find_elements(By.CSS_SELECTOR, ".chat-message.agent"))
+
+    inp = driver.find_element(By.CSS_SELECTOR, "input[name='message']")
+    inp.send_keys(message)
+    inp.send_keys(Keys.RETURN)
+
+    # Wait for a NEW agent response (not one from a prior test)
+    WebDriverWait(driver, 30).until(
+        lambda d: len(d.find_elements(By.CSS_SELECTOR, ".chat-message.agent")) > existing
+    )
+
+    return existing
+
+
+# ── Page load tests ─────────────────────────────────────────────────────────
+
+
+class TestPageLoad:
+    """Dashboard loads and shows expected structure."""
+
+    def test_homepage_loads(self, driver):
+        driver.get(DASHBOARD_URL)
+        assert driver.title != ""
+
+    def test_header_visible(self, driver):
+        _load_dashboard(driver)
+        header = driver.find_element(By.CSS_SELECTOR, ".mc-header, header, nav")
+        assert header.is_displayed()
+
+    def test_sidebar_loads(self, driver):
+        _load_dashboard(driver)
+        _wait_for_sidebar(driver)
+
+    def test_timmy_panel_loads(self, driver):
+        _load_dashboard(driver)
+
+    def test_chat_input_exists(self, driver):
+        _load_dashboard(driver)
+        inp = driver.find_element(By.CSS_SELECTOR, "input[name='message']")
+        assert inp.is_displayed()
+        assert "timmy" in inp.get_attribute("placeholder").lower()
+
+    def test_send_button_exists(self, driver):
+        _load_dashboard(driver)
+        btn = driver.find_element(By.CSS_SELECTOR, "button.mc-btn-send")
+        assert btn.is_displayed()
+        assert "SEND" in btn.text
+
+    def test_health_panel_loads(self, driver):
+        _load_dashboard(driver)
+        WebDriverWait(driver, 10).until(
+            EC.presence_of_element_located(
+                (By.XPATH, "//*[contains(text(), 'SYSTEM HEALTH')]")
+            )
+        )
+
+
+# ── Chat interaction tests ──────────────────────────────────────────────────
+
+
+class TestChatInteraction:
+    """Send a single message and verify all chat-related behaviors at once.
+
+    We only send ONE message to avoid spamming Ollama and crashing the browser.
+    """
+
+    def test_chat_roundtrip(self, driver):
+        """Full chat roundtrip: send message, get response, input clears, chat scrolls."""
+        _load_dashboard(driver)
+
+        # Wait for any initial HTMX requests (history load) to settle
+        time.sleep(2)
+
+        existing_agents = len(driver.find_elements(By.CSS_SELECTOR, ".chat-message.agent"))
+
+        inp = driver.find_element(By.CSS_SELECTOR, "input[name='message']")
+        inp.send_keys("hello from selenium")
+        inp.send_keys(Keys.RETURN)
+
+        # 1. User bubble appears immediately
+        WebDriverWait(driver, 5).until(
+            EC.presence_of_element_located(
+                (By.CSS_SELECTOR, ".chat-message.user")
+            )
+        )
+
+        # 2. Agent response arrives
+        WebDriverWait(driver, 30).until(
+            lambda d: len(d.find_elements(By.CSS_SELECTOR, ".chat-message.agent")) > existing_agents
+        )
+
+        # 3. Input cleared (regression test)
+        time.sleep(0.5)
+        inp = driver.find_element(By.CSS_SELECTOR, "input[name='message']")
+        assert inp.get_attribute("value") == "", "Input should be empty after sending"
+
+        # 4. Chat scrolled to bottom (regression test)
+        chat_log = driver.find_element(By.ID, "chat-log")
+        scroll_top = driver.execute_script("return arguments[0].scrollTop", chat_log)
+        scroll_height = driver.execute_script("return arguments[0].scrollHeight", chat_log)
+        client_height = driver.execute_script("return arguments[0].clientHeight", chat_log)
+
+        if scroll_height > client_height:
+            gap = scroll_height - scroll_top - client_height
+            assert gap < 50, f"Chat not scrolled to bottom (gap: {gap}px)"
+
+
+# ── Task panel tests ────────────────────────────────────────────────────────
+
+
+class TestTaskPanel:
+    """Task creation panel works correctly."""
+
+    def test_task_panel_via_url(self, driver):
+        """Task panel loads correctly when navigated to directly."""
+        driver.get(f"{DASHBOARD_URL}/swarm/tasks/panel")
+        WebDriverWait(driver, 10).until(
+            EC.presence_of_element_located(
+                (By.XPATH, "//*[contains(text(), 'CREATE TASK')]")
+            )
+        )
+
+    def test_task_panel_has_form(self, driver):
+        """Task creation panel has description and agent fields."""
+        driver.get(f"{DASHBOARD_URL}/swarm/tasks/panel")
+        WebDriverWait(driver, 10).until(
+            EC.presence_of_element_located(
+                (By.XPATH, "//*[contains(text(), 'CREATE TASK')]")
+            )
+        )
+
+        driver.find_element(By.CSS_SELECTOR, "textarea[name='description']")
+        driver.find_element(By.CSS_SELECTOR, "select[name='agent_id']")
+
+    def test_task_button_on_agent_card(self, driver):
+        """If agents are registered, TASK button on agent card opens task panel."""
+        _load_dashboard(driver)
+        _wait_for_sidebar(driver)
+
+        if not _has_registered_agents(driver):
+            pytest.skip("No agents registered — TASK button not available")
+
+        task_btn = driver.find_element(
+            By.XPATH,
+            "//div[contains(@class, 'mc-agent-card')]//button[contains(text(), 'TASK')]",
+        )
+        task_btn.click()
+
+        WebDriverWait(driver, 10).until(
+            EC.presence_of_element_located(
+                (By.XPATH, "//*[contains(text(), 'CREATE TASK')]")
+            )
+        )
+
+
+# ── Agent sidebar tests ─────────────────────────────────────────────────────
+
+
+class TestAgentSidebar:
+    """Agent sidebar displays correctly."""
+
+    def test_sidebar_header_shows(self, driver):
+        _load_dashboard(driver)
+        _wait_for_sidebar(driver)
+        header = driver.find_element(
+            By.XPATH, "//*[contains(text(), 'SWARM AGENTS')]"
+        )
+        assert header.is_displayed()
+
+    def test_sidebar_shows_status_when_agents_exist(self, driver):
+        """If agents are registered, cards show status dots."""
+        _load_dashboard(driver)
+        _wait_for_sidebar(driver)
+
+        if not _has_registered_agents(driver):
+            pytest.skip("No agents registered — skipping card test")
+
+        cards = driver.find_elements(By.CSS_SELECTOR, ".mc-agent-card")
+        for card in cards:
+            dots = card.find_elements(By.CSS_SELECTOR, ".status-dot")
+            assert len(dots) >= 1, "Agent card should show a status dot"
+
+    def test_no_agents_fallback(self, driver):
+        """When no agents registered, sidebar shows fallback message."""
+        _load_dashboard(driver)
+        _wait_for_sidebar(driver)
+
+        if _has_registered_agents(driver):
+            pytest.skip("Agents are registered — fallback not shown")
+
+        body = driver.find_element(By.CSS_SELECTOR, ".mc-sidebar").text
+        assert "NO AGENTS REGISTERED" in body
+
+
+# ── Navigation tests ────────────────────────────────────────────────────────
+
+
+class TestNavigation:
+    """Basic navigation flows work end-to-end."""
+
+    def test_clear_chat_button(self, driver):
+        _load_dashboard(driver)
+        clear_btn = driver.find_element(By.XPATH, "//button[contains(text(), 'CLEAR')]")
+        assert clear_btn.is_displayed()
+
+    def test_health_endpoint_returns_200(self, driver):
+        driver.get(f"{DASHBOARD_URL}/health")
+        assert "ok" in driver.page_source
+
+    def test_nav_links_visible(self, driver):
+        _load_dashboard(driver)
+        links = driver.find_elements(By.CSS_SELECTOR, ".mc-desktop-nav .mc-test-link")
+        assert len(links) >= 3, "Navigation should have multiple links"
--- a/tests/test_agent.py
+++ b/tests/test_agent.py
@@ -37,7 +37,9 @@ def test_create_timmy_uses_llama32():
        from timmy.agent import create_timmy
        create_timmy()

-        MockOllama.assert_called_once_with(id="llama3.2")
+        MockOllama.assert_called_once()
+        kwargs = MockOllama.call_args.kwargs
+        assert kwargs["id"] == "llama3.2"


 def test_create_timmy_history_config():
@@ -79,6 +81,46 @@ def test_create_timmy_embeds_system_prompt():
        assert kwargs["description"] == TIMMY_SYSTEM_PROMPT


+# ── Ollama host regression (container connectivity) ─────────────────────────
+
+def test_create_timmy_passes_ollama_url_to_model():
+    """Regression: Ollama model must receive settings.ollama_url as host.
+
+    Without this, containers default to localhost:11434 which is unreachable
+    when Ollama runs on the Docker host.
+    """
+    with patch("timmy.agent.Agent"), \
+         patch("timmy.agent.Ollama") as MockOllama, \
+         patch("timmy.agent.SqliteDb"):
+
+        from timmy.agent import create_timmy
+        create_timmy()
+
+        kwargs = MockOllama.call_args.kwargs
+        assert "host" in kwargs, "Ollama() must receive host= parameter"
+        assert kwargs["host"] == "http://localhost:11434"  # default from config
+
+
+def test_create_timmy_respects_custom_ollama_url():
+    """Ollama host should follow OLLAMA_URL when overridden in config."""
+    custom_url = "http://host.docker.internal:11434"
+    with patch("timmy.agent.Agent"), \
+         patch("timmy.agent.Ollama") as MockOllama, \
+         patch("timmy.agent.SqliteDb"), \
+         patch("timmy.agent.settings") as mock_settings:
+
+        mock_settings.ollama_model = "llama3.2"
+        mock_settings.ollama_url = custom_url
+        mock_settings.timmy_model_backend = "ollama"
+        mock_settings.airllm_model_size = "70b"
+
+        from timmy.agent import create_timmy
+        create_timmy()
+
+        kwargs = MockOllama.call_args.kwargs
+        assert kwargs["host"] == custom_url
+
+
 # ── AirLLM path ──────────────────────────────────────────────────────────────

 def test_create_timmy_airllm_returns_airllm_agent():
--- a/tests/test_swarm_routes_functional.py
+++ b/tests/test_swarm_routes_functional.py
@@ -192,17 +192,11 @@ class TestSwarmUIPartials:
        assert response.status_code == 200
        assert "text/html" in response.headers["content-type"]

-    def test_task_panel_route_shadowed(self, client):
-        """The /swarm/tasks/panel route is shadowed by /swarm/tasks/{task_id}.
-
-        FastAPI matches the dynamic {task_id} route first, so "panel" is
-        treated as a task_id lookup, returning JSON with an error.
-        This documents the current behavior (a routing order issue).
-        """
+    def test_task_panel_route_returns_html(self, client):
+        """The /swarm/tasks/panel route must return HTML, not be shadowed by {task_id}."""
        response = client.get("/swarm/tasks/panel")
        assert response.status_code == 200
-        data = response.json()
-        assert "error" in data
+        assert "text/html" in response.headers["content-type"]

    def test_direct_assign_with_agent(self, client):
        spawn_resp = client.post("/swarm/spawn", data={"name": "Worker"})