diff --git a/test/bds/ai_test.exs b/test/bds/ai_test.exs
index da158de..4395412 100644
--- a/test/bds/ai_test.exs
+++ b/test/bds/ai_test.exs
@@ -361,6 +361,31 @@ defmodule BDS.AITest do
refute Map.has_key?(payload, "tool_choice")
end
+ test "openai-compatible generation disables thinking for configured models" do
+ Application.put_env(:bds, :test_pid, self())
+
+ server =
+ start_supervised!({Bandit, plug: RecordingCompletionServer, port: 0, startup_log: false})
+
+ {:ok, {_address, port}} = ThousandIsland.listener_info(server)
+ assert :ok = BDS.AI.put_model_capabilities("qwen3.5-122b", %{disables_reasoning: true})
+
+ assert {:ok, %{content: "Short Title"}} =
+ BDS.AI.OpenAICompatibleRuntime.generate(
+ %{url: "http://127.0.0.1:#{port}/v1", api_key: nil},
+ %{
+ operation: :chat_title,
+ model: "qwen3.5-122b",
+ messages: [%{"role" => "user", "content" => "Topic: posts per month"}],
+ max_output_tokens: 256
+ },
+ []
+ )
+
+ assert_received {:completion_payload, payload}
+ assert payload["chat_template_kwargs"] == %{"enable_thinking" => false}
+ end
+
test "airplane mode routes title tasks to airplane endpoint and offline title model" do
assert {:ok, _endpoint} =
BDS.AI.put_endpoint(
@@ -506,7 +531,8 @@ defmodule BDS.AITest do
assert :ok =
BDS.AI.put_model_capabilities("llama3.2", %{
supports_attachment: true,
- supports_tool_calls: false
+ supports_tool_calls: false,
+ disables_reasoning: true
})
assert {:ok, analysis} =
@@ -529,6 +555,7 @@ defmodule BDS.AITest do
assert endpoint.kind == :airplane
assert request.operation == :analyze_image
assert request.model == "llama3.2"
+ assert BDS.AI.Catalog.model_capabilities("llama3.2").disables_reasoning
end
test "chat persists user, tool, and assistant messages with usage and blog stats prompt augmentation" do
@@ -702,7 +729,52 @@ defmodule BDS.AITest do
assert_received {:runtime_request, _endpoint, title_request}
assert title_request.operation == :chat_title
assert title_request.model == "title-model"
+ assert title_request.max_output_tokens == 256
assert Enum.any?(title_request.messages, &(&1["content"] =~ "2-3 words"))
+ assert Enum.any?(title_request.messages, &(&1["content"] =~ "Do not include reasoning"))
+ assert Enum.any?(title_request.messages, &(&1["content"] =~ "How many items"))
+ end
+
+ test "chat retries title generation on later turns while the title is still generated" do
+ {:ok, project} = create_project_fixture("Retry Title Chat")
+ _fixtures = seed_project_content(project.id)
+
+ assert {:ok, _endpoint} =
+ BDS.AI.put_endpoint(
+ :online,
+ %{
+ url: "https://api.example.test/v1",
+ api_key: "online-secret",
+ model: "gpt-4o-mini"
+ },
+ secret_backend: FakeSecretBackend
+ )
+
+ assert :ok = BDS.AI.set_airplane_mode(false)
+ assert :ok = BDS.AI.put_model_preference(:title, "title-model")
+ assert {:ok, conversation} = BDS.AI.start_chat(%{title: "New Chat", model: "gpt-4o-mini"})
+
+ Repo.insert!(%BDS.AI.ChatMessage{
+ conversation_id: conversation.id,
+ role: :user,
+ content: "Earlier turn whose title attempt failed",
+ created_at: Persistence.now_ms()
+ })
+
+ assert {:ok, reply} =
+ BDS.AI.send_chat_message(conversation.id, "How many items are in the blog now?",
+ runtime: FakeRuntime,
+ test_pid: self(),
+ project_id: project.id,
+ secret_backend: FakeSecretBackend
+ )
+
+ assert reply.conversation.title == "Blog Stats"
+ assert BDS.AI.get_chat_conversation(conversation.id).title == "Blog Stats"
+
+ assert_received {:runtime_request, _endpoint, %{operation: :chat}}
+ assert_received {:runtime_request, _endpoint, %{operation: :chat}}
+ assert_received {:runtime_request, _endpoint, %{operation: :chat_title} = title_request}
assert Enum.any?(title_request.messages, &(&1["content"] =~ "How many items"))
end
diff --git a/test/bds/desktop/shell_live_test.exs b/test/bds/desktop/shell_live_test.exs
index a33f81c..deb981e 100644
--- a/test/bds/desktop/shell_live_test.exs
+++ b/test/bds/desktop/shell_live_test.exs
@@ -93,6 +93,39 @@ defmodule BDS.Desktop.ShellLiveTest do
end
end
+ defmodule TitleChatServer do
+ use Plug.Router
+ import Phoenix.ConnTest, except: [post: 2]
+
+ plug(:match)
+ plug(:dispatch)
+
+ post "/v1/chat/completions" do
+ {:ok, request_body, conn} = Plug.Conn.read_body(conn)
+ request = Jason.decode!(request_body)
+ send(Application.fetch_env!(:bds, :test_pid), {:title_chat_request, request})
+
+ content =
+ if Enum.any?(request["messages"] || [], fn message ->
+ String.contains?(message["content"] || "", "Generate an ultra-short title")
+ end) do
+ "Posts 2026"
+ else
+ "Ich habe die Posts pro Monat ermittelt."
+ end
+
+ body =
+ Jason.encode!(%{
+ "choices" => [%{"message" => %{"content" => content}}],
+ "usage" => %{"prompt_tokens" => 8, "completion_tokens" => 5}
+ })
+
+ conn
+ |> Plug.Conn.put_resp_content_type("application/json")
+ |> send_resp(200, body)
+ end
+ end
+
@endpoint BDS.Desktop.Endpoint
setup do
@@ -772,6 +805,8 @@ defmodule BDS.Desktop.ShellLiveTest do
assert html =~ "Offline Endpoint URL"
assert html =~ "Online API Key"
assert html =~ "Offline API Key"
+ assert html =~ "Online Chat Reasoning"
+ assert html =~ "Offline Chat Reasoning"
refute html =~ "Mistral API Key"
refute html =~ "Anthropic / Online API Key"
@@ -782,12 +817,14 @@ defmodule BDS.Desktop.ShellLiveTest do
"online_api_key" => "online-secret",
"online_chat_model" => "gpt-4.1",
"online_chat_tools" => "true",
+ "online_chat_disable_reasoning" => "true",
"online_title_model" => "gpt-4.1-mini",
"online_image_analysis_model" => "gpt-4.1-vision",
"offline_url" => "http://localhost:11434/v1",
"offline_api_key" => "",
"offline_chat_model" => "llama3.3",
"offline_chat_tools" => "true",
+ "offline_chat_disable_reasoning" => "true",
"offline_title_model" => "llama3.2",
"offline_image_analysis_model" => "llava:latest",
"offline_mode" => "true",
@@ -816,8 +853,11 @@ defmodule BDS.Desktop.ShellLiveTest do
assert {:ok, "llama3.2"} = AI.get_model_preference(:airplane_title)
assert {:ok, "llava:latest"} = AI.get_model_preference(:airplane_image_analysis)
- assert %{supports_tool_calls: true} = BDS.AI.Catalog.model_capabilities("gpt-4.1")
- assert %{supports_tool_calls: true} = BDS.AI.Catalog.model_capabilities("llama3.3")
+ assert %{supports_tool_calls: true, disables_reasoning: true} =
+ BDS.AI.Catalog.model_capabilities("gpt-4.1")
+
+ assert %{supports_tool_calls: true, disables_reasoning: true} =
+ BDS.AI.Catalog.model_capabilities("llama3.3")
end
test "ai settings refresh models from the configured endpoints" do
@@ -2195,7 +2235,9 @@ defmodule BDS.Desktop.ShellLiveTest do
css = File.read!(Path.expand("../../../priv/ui/app.css", __DIR__))
assert css =~ ".chat-panel-title {"
assert css =~ "overflow: visible;"
- refute css =~ ".chat-panel-title {\n flex: 1;\n min-width: 0;\n display: flex;\n align-items: center;\n gap: 10px;\n overflow: hidden;"
+
+ refute css =~
+ ".chat-panel-title {\n flex: 1;\n min-width: 0;\n display: flex;\n align-items: center;\n gap: 10px;\n overflow: hidden;"
render_click(view, "select_chat_model", %{"model" => "llama-next"})
@@ -2203,6 +2245,70 @@ defmodule BDS.Desktop.ShellLiveTest do
assert render(view) =~ "llama-next"
end
+ test "chat editor updates the visible new-chat title after the first turn" do
+ Application.put_env(:bds, :test_pid, self())
+ assert :ok = AI.set_airplane_mode(false)
+
+ server =
+ start_supervised!({Bandit, plug: TitleChatServer, port: 0, startup_log: false})
+
+ {:ok, {_address, port}} = ThousandIsland.listener_info(server)
+
+ assert {:ok, _endpoint} =
+ AI.put_endpoint(:online, %{
+ url: "http://127.0.0.1:#{port}/v1",
+ api_key: "online-secret",
+ model: "gpt-4.1"
+ })
+
+ assert {:ok, conversation} = AI.start_chat(%{title: "New Chat", model: "gpt-4.1"})
+
+ {:ok, view, _html} = live_isolated(build_conn(), BDS.Desktop.ShellLive)
+ html = render_click(view, "select_view", %{"view" => "chat"})
+
+ assert html =~ ~s(
New Chat)
+
+ html =
+ render_click(view, "pin_sidebar_item", %{
+ "route" => "chat",
+ "id" => conversation.id,
+ "title" => conversation.title,
+ "subtitle" => conversation.model || "chat"
+ })
+
+ assert html =~ ~s(
New Chat)
+
+ _html =
+ render_change(view, "change_chat_editor_input", %{"message" => "Posts pro Monat 2026"})
+
+ _html =
+ view
+ |> element("[data-testid='chat-send-button']")
+ |> render_click()
+
+ Process.sleep(350)
+ html = render(view)
+
+ assert_received {:title_chat_request, chat_request}
+
+ refute Enum.any?(chat_request["messages"] || [], fn message ->
+ String.contains?(message["content"] || "", "Generate an ultra-short title")
+ end)
+
+ assert_received {:title_chat_request, title_request}
+
+ assert Enum.any?(title_request["messages"] || [], fn message ->
+ String.contains?(message["content"] || "", "Generate an ultra-short title")
+ end)
+
+ assert AI.get_chat_conversation(conversation.id).title == "Posts 2026"
+ assert html =~ ~s(
Posts 2026)
+ assert html =~ ~r/
\s*Posts 2026\s*<\/span>/
+ assert html =~ ~s(Posts 2026)
+ refute html =~ ~s(New Chat)
+ refute html =~ ~s(New Chat)
+ end
+
test "chat editor renders legacy model controls, collapsed tool pills, and dismissible A2UI surfaces" do
assert {:ok, conversation} = AI.start_chat(%{title: "Editor Chat", model: "gpt-4.1"})
@@ -2433,7 +2539,9 @@ defmodule BDS.Desktop.ShellLiveTest do
test "chat editor hook reopens server-expanded A2UI surfaces after patches" do
live_js = File.read!(Path.expand("../../../priv/ui/live.js", __DIR__))
- chat_editor = File.read!(Path.expand("../../../lib/bds/desktop/shell_live/chat_editor.ex", __DIR__))
+
+ chat_editor =
+ File.read!(Path.expand("../../../lib/bds/desktop/shell_live/chat_editor.ex", __DIR__))
assert chat_editor =~ "data-expanded={surface_expanded_attr(@surface)}"
assert live_js =~ "this.syncExpandedSurfaces = () =>"
@@ -2508,7 +2616,8 @@ defmodule BDS.Desktop.ShellLiveTest do
assert html =~ "Here is the chart."
assert html =~ ~s(Assistant)
- assert length(:binary.matches(html, ~s(Assistant))) == 1
+ assert length(:binary.matches(html, ~s(Assistant))) ==
+ 1
end
test "chat editor marks user message text as compact" do