v2.18.4 (#866)

Co-authored-by: Albert Jiang <aj@mistral.ai>
Co-authored-by: Clément Drouin <clement.drouin@mistral.ai>
Co-authored-by: Laure Hugo <201583486+laure0303@users.noreply.github.com>
Co-authored-by: Mathias Gesbert <mathias.gesbert@mistral.ai>
Co-authored-by: Mert Unsal <mert.unsal@mistral.ai>
Co-authored-by: Michel Thomazo <51709227+michelTho@users.noreply.github.com>
Co-authored-by: Paul VEZIA <166131032+le-codeur-rapide@users.noreply.github.com>
Co-authored-by: Pierre Rossinès <pierre.rossines@mistral.ai>
Co-authored-by: Quentin <quentin.torroba@mistral.ai>
Co-authored-by: Mistral Vibe <vibe@mistral.ai>

tests/cli/test_help.py

@@ -17,7 +17,8 @@ def test_help_shows_auto_approve_flag(
     output = capsys.readouterr().out
     assert "--auto-approve" in output
     assert "--yolo" in output
-    assert "Shortcut for --agent auto-approve" in output
+    assert "Approves all tool calls without prompting" in output
+    assert "selected agent" in output
 
 
 def test_help_shows_check_upgrade_flag(
@@ -47,13 +48,12 @@ def test_yolo_alias_selects_auto_approve(monkeypatch: pytest.MonkeyPatch) -> Non
 
 
 @pytest.mark.parametrize("flag", ["--auto-approve", "--yolo"])
-def test_auto_approve_aliases_conflict_with_agent(
-    flag: str, monkeypatch: pytest.MonkeyPatch, capsys: pytest.CaptureFixture[str]
+def test_auto_approve_aliases_can_be_combined_with_agent(
+    flag: str, monkeypatch: pytest.MonkeyPatch
 ) -> None:
-    monkeypatch.setattr("sys.argv", ["vibe", "--agent", "plan", flag])
+    monkeypatch.setattr("sys.argv", ["vibe", "--agent", "lean", flag])
 
-    with pytest.raises(SystemExit) as exc_info:
-        parse_arguments()
+    args = parse_arguments()
 
-    assert exc_info.value.code == 2
-    assert "not allowed with argument --agent" in capsys.readouterr().err
+    assert args.agent == "lean"
+    assert args.auto_approve is True

tests/cli/test_initial_agent_name.py

@@ -55,8 +55,15 @@ def test_programmatic_mode_keeps_explicit_agent_arg() -> None:
     assert get_initial_agent_name(args, config) == "accept-edits"
 
 
-def test_auto_approve_flag_selects_auto_approve_agent() -> None:
+def test_auto_approve_flag_keeps_config_default_agent() -> None:
     config = VibeConfig.model_construct(default_agent=BuiltinAgentName.PLAN)
     args = _make_args(agent=None, prompt="hello", auto_approve=True)
 
-    assert get_initial_agent_name(args, config) == BuiltinAgentName.AUTO_APPROVE
+    assert get_initial_agent_name(args, config) == BuiltinAgentName.PLAN
+
+
+def test_auto_approve_flag_keeps_explicit_agent_arg() -> None:
+    config = VibeConfig.model_construct(default_agent=BuiltinAgentName.PLAN)
+    args = _make_args(agent="lean", prompt="hello", auto_approve=True)
+
+    assert get_initial_agent_name(args, config) == BuiltinAgentName.LEAN

tests/cli/test_programmatic_setup.py

@@ -22,6 +22,7 @@ def _make_args(**overrides: object) -> argparse.Namespace:
         "enabled_tools": None,
         "output": "text",
         "agent": "default",
+        "auto_approve": False,
         "check_upgrade": False,
         "setup": False,
         "workdir": None,
@@ -301,6 +302,35 @@ def test_run_cli_passes_max_tokens_to_run_programmatic(
     assert call["max_session_tokens"] == 123
 
 
+def test_run_cli_auto_approve_sets_config_without_changing_agent(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    args = _make_args(agent="lean", auto_approve=True)
+    call: dict[str, object] = {}
+    config = build_test_vibe_config(default_agent="plan")
+
+    monkeypatch.setattr(cli_mod, "bootstrap_config_files", lambda: None)
+    monkeypatch.setattr(cli_mod, "load_config_or_exit", lambda interactive: config)
+    monkeypatch.setattr(cli_mod, "load_hooks_from_fs", lambda _config: None)
+    monkeypatch.setattr(cli_mod, "setup_tracing", lambda _config: None)
+    monkeypatch.setattr(cli_mod, "load_session", lambda _args, _config: None)
+    monkeypatch.setattr(cli_mod, "get_prompt_from_stdin", lambda: None)
+    monkeypatch.setattr(cli_mod, "warn_if_workdir_trust_is_unset", lambda: None)
+
+    def fake_run_programmatic(**kwargs: object) -> str:
+        call.update(kwargs)
+        return "done"
+
+    monkeypatch.setattr(cli_mod, "run_programmatic", fake_run_programmatic)
+
+    with pytest.raises(SystemExit) as exc_info:
+        cli_mod.run_cli(args)
+
+    assert exc_info.value.code == 0
+    assert call["agent_name"] == "lean"
+    assert config.bypass_tool_permissions is True
+
+
 def test_run_cli_runs_update_prompt_before_trust_resolver(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:

tests/cli/test_terminal_detect.py

@@ -37,6 +37,7 @@ def test_detects_cursor_from_vscode_environment() -> None:
 @pytest.mark.parametrize(
     ("term_program", "terminal"),
     [
+        ("Apple_Terminal", Terminal.APPLE_TERMINAL),
         ("iterm.app", Terminal.ITERM2),
         ("wezterm", Terminal.WEZTERM),
         ("ghostty", Terminal.GHOSTTY),
@@ -58,6 +59,7 @@ def test_detects_term_program_mapping(term_program: str, terminal: Terminal) ->
         ("ALACRITTY_SOCKET", Terminal.ALACRITTY),
         ("ALACRITTY_LOG", Terminal.ALACRITTY),
         ("WT_SESSION", Terminal.WINDOWS_TERMINAL),
+        ("WT_PROFILE_ID", Terminal.WINDOWS_TERMINAL),
     ],
 )
 def test_detects_environment_marker_fallback(env_var: str, terminal: Terminal) -> None:

tests/cli/test_ui_teleport_availability.py

@@ -9,12 +9,14 @@ import pytest
 from tests.cli.plan_offer.adapters.fake_whoami_gateway import FakeWhoAmIGateway
 from tests.conftest import build_test_vibe_app, build_test_vibe_config
 from tests.constants import OPENAI_BASE_URL
+from vibe import __version__
 from vibe.cli.plan_offer.ports.whoami_gateway import WhoAmIPlanType, WhoAmIResponse
 from vibe.cli.textual_ui.widgets.chat_input import ChatInputContainer
 from vibe.cli.textual_ui.widgets.messages import ErrorMessage
 from vibe.cli.textual_ui.widgets.no_markup_static import NoMarkupStatic
 from vibe.core.config import ModelConfig, ProviderConfig, VibeConfig
 from vibe.core.types import Backend
+from vibe.core.utils import get_platform_id, get_platform_version
 
 
 def _chat_plan_gateway(*, prompt_switching_to_pro_plan: bool) -> FakeWhoAmIGateway:
@@ -40,6 +42,13 @@ async def _wait_until(pause, predicate, timeout: float = 2.0) -> None:
     raise AssertionError("Condition was not met within the timeout")
 
 
+def _expected_system_metadata() -> dict[str, Any]:
+    metadata: dict[str, Any] = {"os": get_platform_id(), "version": __version__}
+    if os_version := get_platform_version():
+        metadata["os_version"] = os_version
+    return metadata
+
+
 def _teleport_failed_events(
     telemetry_events: list[dict[str, Any]],
 ) -> list[dict[str, Any]]:
@@ -114,6 +123,7 @@ async def test_teleport_command_without_history_sends_early_failure_telemetry(
         {
             "event_name": "vibe.teleport_failed",
             "properties": {
+                **_expected_system_metadata(),
                 "stage": "no_history",
                 "error_class": "TeleportNoHistoryError",
                 "push_required": False,
@@ -156,6 +166,7 @@ async def test_teleport_command_visible_but_errors_when_key_not_eligible(
         {
             "event_name": "vibe.teleport_failed",
             "properties": {
+                **_expected_system_metadata(),
                 "stage": "ineligible",
                 "error_class": "TeleportIneligibleError",
                 "push_required": False,

tests/cli/textual_ui/test_diff_rendering.py

@@ -1,12 +1,17 @@
 from __future__ import annotations
 
+from pathlib import Path
+
+import pytest
 from textual.content import Content
 from textual.highlight import HighlightTheme
 from textual.widget import Widget
 
 from vibe.cli.textual_ui.widgets.diff_rendering import (
+    DiffOccurrence,
     _build_diff_line,
     diff_border_colors,
+    edit_diff_inputs,
     language_for_path,
     render_edit_diff,
 )
@@ -20,14 +25,12 @@ def _build(
     )
 
 
-def _render(*args, **kwargs):
-    kwargs.setdefault("dark", True)
-    args = list(args)
-    # Tests pass a single start line as an int for readability; the renderer
-    # expects a list of occurrences.
-    if len(args) >= 4 and isinstance(args[3], int):
-        args[3] = [args[3]]
-    return render_edit_diff(*args, **kwargs)
+def _render(old_string, new_string, language, start, *, ansi, dark=True):
+    # Tests pass the same old/new at a single start line (int/None) or, for
+    # replace_all, at a list of start lines; build one occurrence per location.
+    starts = start if isinstance(start, list) else [start]
+    occurrences = [DiffOccurrence(s, old_string, new_string) for s in starts]
+    return render_edit_diff(occurrences, language, ansi=ansi, dark=dark)
 
 
 def _render_with_colors(*args, **kwargs):
@@ -80,14 +83,20 @@ class TestBuildDiffLine:
         styles = _styles_at(content, 0)
         assert "$text-success" in styles
         assert all("dim" not in s for s in styles)
+        assert all("bold" not in s for s in styles)
 
-    def test_removed_line_number_and_sign_bright_in_ansi(self) -> None:
+    def test_removed_line_number_not_bold_in_ansi(self) -> None:
         content = _build("x = 1", "-", 10, "py", ansi=True)
         lineno_styles = _styles_at(content, 0)
-        sign_styles = _styles_at(content, 5)
-        assert any("bold" in s and "$text-error" in s for s in lineno_styles)
-        assert any("bold" in s and "$text-error" in s for s in sign_styles)
+        assert any("$text-error" in s for s in lineno_styles)
+        assert all("bold" not in s for s in lineno_styles)
         assert all("dim" not in s for s in lineno_styles)
+
+    def test_removed_sign_not_bold_in_ansi(self) -> None:
+        content = _build("x = 1", "-", 10, "py", ansi=True)
+        sign_styles = _styles_at(content, 5)
+        assert any("$text-error" in s for s in sign_styles)
+        assert all("bold" not in s for s in sign_styles)
         assert all("dim" not in s for s in sign_styles)
 
     def test_line_number_dimmed_for_unchanged_rows_in_ansi(self) -> None:
@@ -139,6 +148,26 @@ class TestRenderEditDiff:
         widgets = _render("x = 100", "x = 200", "py", 42, ansi=False)
         assert any("42" in _plain(w) for w in widgets)
 
+    @pytest.mark.asyncio
+    async def test_leading_newline_snippet_gutter_matches_file_lines(
+        self, tmp_path: Path
+    ) -> None:
+        # A leading-newline snippet starts modifying the previous line; the
+        # whole-line diff must show that line and number the hunk against the
+        # real file lines, not skip the leading newline and drift by one.
+        f = tmp_path / "f.py"
+        f.write_text("aa\nbab\ncc\n")
+        occurrences = await edit_diff_inputs(
+            str(f), "\nbab\nc", "Z\nbab\nC", replace_all=False
+        )
+        widgets = render_edit_diff(occurrences, "py", ansi=False, dark=True)
+        file_lines = ["aa", "bab", "cc"]
+        for w in widgets:
+            plain = _plain(w)
+            if plain[5:6] in (" ", "-"):
+                lineno = int(plain[:4])
+                assert plain[7:] == file_lines[lineno - 1]
+
     def test_multi_hunk_line_numbers(self) -> None:
         search = "A\nB\nC\nD\nE\nF\nG\nH"
         replace = "Z\nB\nC\nD\nE\nF\nG\nY"
@@ -164,31 +193,45 @@ class TestRenderEditDiff:
         assert any(_plain(w).rstrip().endswith("Z") for w in added)
 
     def test_replace_all_renders_each_occurrence(self) -> None:
-        widgets = render_edit_diff(
-            "foo", "bar", "py", [3, 10, 25], ansi=False, dark=True
-        )
+        widgets = _render("foo", "bar", "py", [3, 10, 25], ansi=False)
         removed = [w for w in widgets if "diff-removed" in w.classes]
         added = [w for w in widgets if "diff-added" in w.classes]
         assert len(removed) == 3
         assert len(added) == 3
 
     def test_replace_all_uses_each_start_line(self) -> None:
-        widgets = render_edit_diff(
-            "foo", "bar", "py", [3, 10, 25], ansi=False, dark=True
-        )
+        widgets = _render("foo", "bar", "py", [3, 10, 25], ansi=False)
         joined = "\n".join(_plain(w) for w in widgets)
         assert "3" in joined
         assert "10" in joined
         assert "25" in joined
 
     def test_replace_all_separates_occurrences_with_gap(self) -> None:
-        widgets = render_edit_diff("foo", "bar", "py", [3, 10], ansi=False, dark=True)
+        widgets = _render("foo", "bar", "py", [3, 10], ansi=False)
         assert sum("diff-gap" in w.classes for w in widgets) == 1
 
     def test_single_occurrence_has_no_gap(self) -> None:
-        widgets = render_edit_diff("foo", "bar", "py", [3], ansi=False, dark=True)
+        widgets = _render("foo", "bar", "py", [3], ansi=False)
         assert all("diff-gap" not in w.classes for w in widgets)
 
+    def test_per_occurrence_full_lines(self) -> None:
+        # Each occurrence carries its own whole-line content with its own line.
+        widgets = render_edit_diff(
+            [
+                DiffOccurrence(2, "x = bar + 1", "x = qux + 1"),
+                DiffOccurrence(7, "y = bar - 2", "y = qux - 2"),
+            ],
+            "py",
+            ansi=False,
+            dark=True,
+        )
+        removed = [_plain(w) for w in widgets if "diff-removed" in w.classes]
+        added = [_plain(w) for w in widgets if "diff-added" in w.classes]
+        assert any("x = bar + 1" in r for r in removed)
+        assert any("y = bar - 2" in r for r in removed)
+        assert any("x = qux + 1" in a for a in added)
+        assert any("y = qux - 2" in a for a in added)
+
 
 class TestBorderColors:
     def test_keys_index_into_widgets(self) -> None:

tests/cli/textual_ui/test_tool_error_muting_widgets.py

@@ -49,12 +49,23 @@ def _call_event() -> ToolCallEvent:
     )
 
 
-def _error_result() -> ToolResultEvent:
+def _error_result(error: str = "boom") -> ToolResultEvent:
     return ToolResultEvent(
         tool_name="stub_tool",
         tool_class=FakeTool,
         result=None,
-        error="boom",
+        error=error,
+        tool_call_id="a",
+    )
+
+
+def _skipped_result() -> ToolResultEvent:
+    return ToolResultEvent(
+        tool_name="stub_tool",
+        tool_class=FakeTool,
+        result=None,
+        skipped=True,
+        skip_reason="User declined",
         tool_call_id="a",
     )
 
@@ -87,6 +98,40 @@ async def test_error_renders_muted_then_escalates_icon_and_styling() -> None:
         assert not result_widget.has_class("error-text")
 
 
+@pytest.mark.asyncio
+async def test_declined_call_renders_muted_square() -> None:
+    app = _ToolApp(_call_event(), _skipped_result())
+    async with app.run_test() as pilot:
+        await pilot.pause()
+        call_widget = app.call_widget
+        assert call_widget is not None
+
+        icon = call_widget._indicator_widget
+        assert icon is not None
+        assert _rendered(icon).plain == "□"
+        assert icon.has_class("muted")
+        assert not icon.has_class("error")
+
+
+@pytest.mark.asyncio
+async def test_error_with_square_brackets_does_not_raise_markup_error() -> None:
+    error = (
+        "Validation error in tool ask_user_question: 1 validation error for "
+        "AskUserQuestionArgs\nquestions.0.header\n  Value error "
+        "[type=value_error, input_value={'questions[0].header': 'x'}, "
+        "input_type=dict]"
+    )
+    app = _ToolApp(_call_event(), _error_result(error))
+    async with app.run_test() as pilot:
+        await pilot.pause()
+        result_widget = app.result_widget
+        assert result_widget is not None
+
+        detail = result_widget.query_one(CollapsibleSection).query_one(Static)
+        content = _rendered(detail)
+        assert content.plain == f"Error: {error}"
+
+
 @pytest.mark.asyncio
 async def test_folded_error_detail_colors_only_the_error_word() -> None:
     app = _ToolApp(_call_event(), _error_result())