Skip to content

armgit / avibe

Public Imported from GitHub
0

Make your computer a home for your AI partner. Avibe is the local-first Agent OS: Vibey, its built-in AI partner, works out of the box, and Claude Code, Codex & OpenCode run unchanged beside it — reach them from your browser, phone, or any chat app.

fix(vibey): say Vibey cannot run where no gateway engine is built (#2454)

Call Whiteb2d631572dc5
diff --git a/config/v2_config.py b/config/v2_config.py
index 73376a98e..68e401326 100644
--- a/config/v2_config.py
+++ b/config/v2_config.py
@@ -1603,19 +1603,25 @@ def is_model_hub_enabled(environ: Optional[Mapping[str, str]] = None) -> bool:
return source.get(MODEL_HUB_ENABLED_ENV, "1").strip().lower() in {"1", "true", "yes", "on"}
-HubSupplyBlock = Literal["hub_disabled", "gateway_off"]
+HubSupplyBlock = Literal["hub_disabled", "engine_unsupported", "gateway_off"]
def hub_supply_block(backend: str, hub: Optional["ModelHubConfig"]) -> Optional[HubSupplyBlock]:
"""Why the Model Hub leaves ``backend`` no model to run on now, or ``None``: the one owner of that answer.
A backend without its own CLI gets every model through the Hub, so the operator's
- Model Hub switch (``VIBE_MODEL_HUB_ENABLED``) leaves it none. Any backend the Hub
- supplies has none while the user has turned the gateway off in Models.
+ Model Hub switch (``VIBE_MODEL_HUB_ENABLED``) leaves it none, and so does a host the
+ pinned engine manifest has no asset for. Any backend the Hub supplies has none while
+ the user has turned the gateway off in Models.
"""
if not is_model_hub_enabled():
return None if backend in NATIVE_CLI_BACKENDS else "hub_disabled"
+ if backend not in NATIVE_CLI_BACKENDS:
+ from vibe.model_hub_runtime.installer import EngineRuntimeManager
+
+ if not EngineRuntimeManager(offline=True).supports_host_platform():
+ return "engine_unsupported"
if hub is not None and not hub.enabled and hub.agents[backend].mode == "hub":
return "gateway_off"
return None
diff --git a/docs/plans/avibe-agent-always-on.md b/docs/plans/avibe-agent-always-on.md
index 988a173ec..41ca6e759 100644
--- a/docs/plans/avibe-agent-always-on.md
+++ b/docs/plans/avibe-agent-always-on.md
@@ -45,12 +45,18 @@ backend list, it is always enabled, and it cannot be turned off.
- **Model Hub cannot supply.** `hub_supply_block(backend, hub_config)` (`config/v2_config.py`) is the one owner of why
the Model Hub leaves a backend no model to run on now. The shared model gate and the adapter's preflight render its
copy (`hub_supply_refusal`), `get_backend_connection` reports such a backend unready, and `GET /api/config` carries
- it as `agent_supply_blocks`, which the Backends pages show. The backend stays listed and enabled in both cases.
+ it as `agent_supply_blocks`, which the Backends pages show. The backend stays listed and enabled in every case.
- Gateway off (Models → Start/Stop): its row says "Gateway off", its page links to Models, and a Turn refuses with
`errors.modelGatewayOff`.
- Model Hub disabled on the instance (`VIBE_MODEL_HUB_ENABLED=0`): a backend without its own CLI gets every model
through the Hub, so its row says "Model Hub disabled", its page says so without a Models link, and a Turn refuses
with `errors.modelHubDisabled`.
+ - No engine for this host (`engine_unsupported`): the pinned engine manifest has no asset for the host's platform
+ (`EngineRuntimeManager.supports_host_platform()`; native Windows today), so a backend without its own CLI has no
+ gateway to run through. Its row and wizard card say "No gateway component for this platform yet", the wizard does
+ not offer it as the entry, and a Turn refuses with `errors.modelGatewayUnsupported`. The manifest is the only
+ input: a platform gains the backend when its engine asset ships. Native backends keep their own Direct path and
+ are unchanged.
## Known by design
diff --git a/docs/plans/setup-three-screen/contracts.md b/docs/plans/setup-three-screen/contracts.md
index 0ed33fc20..8d79fe8b4 100644
--- a/docs/plans/setup-three-screen/contracts.md
+++ b/docs/plans/setup-three-screen/contracts.md
@@ -648,7 +648,7 @@ the default journey. Read failures authorize neither lifecycle writes nor comple
| route control | Confirmed Hub cards with a designated builtin Agent model show that model's first chain hop and backup count. Enabled cards open `RouteChainDialog` for only `(backend, menu model)`; disabled cards preview the same route but cannot edit it. A current successful route read is required to open the editor; pending or failed reads do not infer an empty model or authorize editing. Route-read failure has its own Retry. A CLI assistant's Hub card with no model shows a hint to repair it later in Agent/Model Hub settings; Setup does not choose or write that Agent's model. The built-in card is the D3 exception (C4). Direct cards retain native connection actions, but Direct cannot satisfy setup completion. Unknown or conflicting Hub/Direct ownership disables configuration until a fresh read. A disabled Hub configuration retains C2 recovery, never a Direct setup-completion bypass |
| candidate identity | for each backend, read its current designated builtin Agent, including disabled Agents, and use its saved menu model; custom Agents or the global default do not substitute as that card's route target. C4 may preserve a runnable custom default independently |
| route close | close/commit refreshes the card's server-backed route and connection; `RouteChainDialog` owns the edit draft and save lifecycle, not `SetupFlowState` |
-| built-in card | first in the lineup, no switch and no lifecycle: an always-on badge, its route as above once its Agent has a model, otherwise the inline model offer of C4's D3 override. A Hub supply block (`agent_supply_blocks`: gateway off, Hub disabled) is stated on the card and holds its model row still; the shell keeps its own gateway alert and Retry |
+| built-in card | first in the lineup, no switch and no lifecycle: an always-on badge, its route as above once its Agent has a model, otherwise the inline model offer of C4's D3 override. A Hub supply block (`agent_supply_blocks`: gateway off, Hub disabled, no engine for this platform) is stated on the card and holds its model row still; the shell keeps its own gateway alert and Retry |
| all uninstalled | one install action per CLI assistant, Back still available; workspace entry stays disabled unless the built-in card is ready |
### Per-assistant Model Hub route mapping
diff --git a/modules/agents/model_hub.py b/modules/agents/model_hub.py
index dedbee060..95d1b6093 100644
--- a/modules/agents/model_hub.py
+++ b/modules/agents/model_hub.py
@@ -364,7 +364,11 @@ def launch_refusal_copy(controller: Any, error: BaseException) -> str | None:
return render_turn_outcome_copy(outcome, _language(controller))
-_HUB_SUPPLY_REFUSALS = {"hub_disabled": "errors.modelHubDisabled", "gateway_off": "errors.modelGatewayOff"}
+_HUB_SUPPLY_REFUSALS = {
+ "hub_disabled": "errors.modelHubDisabled",
+ "engine_unsupported": "errors.modelGatewayUnsupported",
+ "gateway_off": "errors.modelGatewayOff",
+}
def hub_supply_refusal(controller: Any, backend: str, language: str) -> str | None:
diff --git a/tests/test_agent_backend_catalog.py b/tests/test_agent_backend_catalog.py
index 996137642..7935f8927 100644
--- a/tests/test_agent_backend_catalog.py
+++ b/tests/test_agent_backend_catalog.py
@@ -77,7 +77,7 @@ def test_copy_about_the_built_in_names_it_from_the_catalog(lang) -> None:
for kind in ("aborted", "UnsupportedModelRoute", "ProviderProtocolViolation", None):
text = error_text(kind, lang)
assert name in text and "{backend}" not in text
- for key in ("errors.modelHubDisabled", "errors.modelGatewayOff"):
+ for key in ("errors.modelHubDisabled", "errors.modelGatewayUnsupported", "errors.modelGatewayOff"):
text = t(key, lang, backend=name)
assert name in text and "{backend}" not in text
always_on = t("error.agentLifecycle.agent_always_enabled.message", lang, agent="vibey")
diff --git a/tests/test_backend_connection.py b/tests/test_backend_connection.py
index de2323d23..4ef3a4127 100644
--- a/tests/test_backend_connection.py
+++ b/tests/test_backend_connection.py
@@ -435,7 +435,7 @@ def _store_hub_credential(source):
def test_vibey_connection_uses_hub_supply_without_any_native_probe(hub_connection, monkeypatch):
"""The new backend has no CLI; the existing native cases cannot protect that boundary."""
- from core import backend_restart
+ from core import backend_restart, managed_runtime
def forbidden(*_args, **_kwargs):
pytest.fail("Avibe connection inspected a native CLI or takeover journal")
@@ -465,6 +465,13 @@ def test_vibey_connection_uses_hub_supply_without_any_native_probe(hub_connectio
stopped = asyncio.run(api.get_backend_connection("vibey"))
assert stopped["enabled"] and stopped["auth"] == "none"
assert not stopped["ready"] and not stopped["entry_eligible"]
+ # A host the pinned engine manifest has no asset for leaves it nothing either, gateway on or off.
+ hub_connection.config.model_hub.enabled = True
+ hub_connection.config.save()
+ monkeypatch.setattr(managed_runtime, "runtime_platform_tag", lambda: "win32-x64")
+ unsupported = asyncio.run(api.get_backend_connection("vibey"))
+ assert unsupported["enabled"] and unsupported["auth"] == "none"
+ assert not unsupported["ready"] and not unsupported["entry_eligible"]
@pytest.mark.parametrize("backend", ["claude", "codex", "opencode"])
diff --git a/tests/test_ui_server_fastapi.py b/tests/test_ui_server_fastapi.py
index 006e6c8c8..32451871c 100644
--- a/tests/test_ui_server_fastapi.py
+++ b/tests/test_ui_server_fastapi.py
@@ -2,6 +2,7 @@ import asyncio
import gzip
import json
import threading
+from pathlib import Path
from types import SimpleNamespace
import pytest
@@ -2097,24 +2098,48 @@ def test_config_routes_redact_platform_and_gateway_secrets(monkeypatch, tmp_path
@pytest.mark.parametrize(
- ("hub_switch", "gateway_enabled", "blocks"),
- [("1", True, {}), ("1", False, {"vibey": "gateway_off"}), ("0", True, {"vibey": "hub_disabled"})],
+ ("hub_switch", "gateway_enabled", "host", "blocks"),
+ [
+ ("1", True, "supported", {}),
+ ("1", False, "supported", {"vibey": "gateway_off"}),
+ ("0", True, "supported", {"vibey": "hub_disabled"}),
+ # No engine asset for the host: the pinned manifest alone decides, on any OS.
+ ("1", True, "win32-x64", {"vibey": "engine_unsupported"}),
+ ("1", True, "manifest_without_host", {"vibey": "engine_unsupported"}),
+ ("1", False, "win32-x64", {"vibey": "engine_unsupported"}),
+ ("0", True, "win32-x64", {"vibey": "hub_disabled"}),
+ ],
)
-def test_config_api_carries_why_the_model_hub_leaves_a_backend_no_model(monkeypatch, tmp_path, hub_switch, gateway_enabled, blocks):
+def test_config_api_carries_why_the_model_hub_leaves_a_backend_no_model(
+ monkeypatch, tmp_path, hub_switch, gateway_enabled, host, blocks
+):
"""The Backends page shows the server's answer; it keeps no rule of its own."""
+ from core import managed_runtime
from modules.agents.catalog import NATIVE_CLI_BACKENDS
+ from vibe.model_hub_runtime.installer import EngineRuntimeManager
monkeypatch.setenv("AVIBE_HOME", str(tmp_path))
monkeypatch.setenv("VIBE_MODEL_HUB_ENABLED", hub_switch)
+ if host == "win32-x64":
+ monkeypatch.setattr(managed_runtime, "runtime_platform_tag", lambda: host)
+ elif host == "manifest_without_host":
+ from vibe.model_hub_runtime import installer
+
+ packaged = json.loads((Path(installer.__file__).parent / "cliproxyapi_manifest.json").read_text())
+ own = EngineRuntimeManager().host_platform()
+ manifest = tmp_path / "engine-manifest.json"
+ manifest.write_text(json.dumps({**packaged, "assets": [a for a in packaged["assets"] if a["platform"] != own]}))
+ monkeypatch.setenv("VIBE_MODEL_HUB_ENGINE_MANIFEST_PATH", str(manifest))
from vibe import api
config = api.save_config(_full_config_payload())
# The gateway turns off only while no native backend is in Hub mode (``runtime_in_use``).
for backend in NATIVE_CLI_BACKENDS:
- config.model_hub.agents[backend].mode = "direct"
+ config.model_hub.agents[backend].mode = "hub" if gateway_enabled else "direct"
config.model_hub.enabled = gateway_enabled
config.save()
+ # A native backend keeps its own Direct path, so a missing engine never blocks it.
assert app.test_client().get("/api/config").get_json()["agent_supply_blocks"] == blocks
diff --git a/tests/test_vibey_agent.py b/tests/test_vibey_agent.py
index fde3029ef..7b9d921d4 100644
--- a/tests/test_vibey_agent.py
+++ b/tests/test_vibey_agent.py
@@ -1431,11 +1431,19 @@ async def test_a_failure_while_settling_reaches_the_shared_owner(engine, session
await harness.agent.handle_message(harness.request("hello"))
[email protected](("hub_switch", "gateway_enabled"), [("1", True), ("1", False), ("0", True)])
+ ("hub_switch", "gateway_enabled", "host"),
+ [("1", True, None), ("1", False, None), ("0", True, None), ("1", True, "win32-x64")],
+)
async def test_a_refused_model_route_fails_before_the_input_is_written(
- engine, session, tmp_path, published, monkeypatch, hub_switch, gateway_enabled
+ engine, session, tmp_path, published, monkeypatch, hub_switch, gateway_enabled, host
) -> None:
+ from core import managed_runtime
+
monkeypatch.setenv("VIBE_MODEL_HUB_ENABLED", hub_switch)
+ if host is not None:
+ # A host the pinned engine manifest has no asset for, simulated on any OS.
+ monkeypatch.setattr(managed_runtime, "runtime_platform_tag", lambda: host)
harness = _Harness(engine, tmp_path, "telegram", [])
async def refuse(*_args, **_kwargs):
@@ -1450,9 +1458,12 @@ async def test_a_refused_model_route_fails_before_the_input_is_written(
assert harness.controller.started == []
assert await harness.context_rows() == []
assert harness.provider.requests == []
- # Every route runs through the Model Hub: disabled, or its gateway off, the copy names that cause.
+ # Every route runs through the Model Hub: disabled, without an engine for this host, or its
+ # gateway off, the copy names that cause instead of the engine error.
if hub_switch == "0":
expected = i18n_t("errors.modelHubDisabled", "en", backend="Vibey")
+ elif host is not None:
+ expected = i18n_t("errors.modelGatewayUnsupported", "en", backend="Vibey")
elif not gateway_enabled:
expected = i18n_t("errors.modelGatewayOff", "en", backend="Vibey")
else:
diff --git a/ui/src/components/onboarding/BuiltinAssistant.test.tsx b/ui/src/components/onboarding/BuiltinAssistant.test.tsx
index 29d4a3c6f..bd8517b67 100644
--- a/ui/src/components/onboarding/BuiltinAssistant.test.tsx
+++ b/ui/src/components/onboarding/BuiltinAssistant.test.tsx
@@ -271,4 +271,16 @@ describe('the built-in assistant card', () => {
await act(async () => undefined);
expect(enter().disabled).toBe(true);
});
+
+ // A host the pinned gateway engine has no build for: Vibey stays listed and always on,
+ // but the card says why it cannot run and is never offered as the way in.
+ it('says this platform has no gateway component and never offers it as the entry', async () => {
+ render(wrap(<AgentDetection data={data({ agent_supply_blocks: { vibey: 'engine_unsupported' } })}
+ onNext={vi.fn()} onNavigate={vi.fn()} agentReads={reads} />));
+ expect(await card().findByText(en.onboarding.setup.noteGatewayUnsupported.replace('{{name}}', 'Vibey'))).toBeTruthy();
+ expect(card().getByText(en.onboarding.setup.builtinAlwaysOn)).toBeTruthy();
+ expect(card().queryByRole('group')).toBeNull();
+ await act(async () => undefined);
+ expect(enter().disabled).toBe(true);
+ });
});
diff --git a/ui/src/components/onboarding/BuiltinAssistantRow.tsx b/ui/src/components/onboarding/BuiltinAssistantRow.tsx
index b65b3de24..3fb3eecfc 100644
--- a/ui/src/components/onboarding/BuiltinAssistantRow.tsx
+++ b/ui/src/components/onboarding/BuiltinAssistantRow.tsx
@@ -57,6 +57,7 @@ export function BuiltinAssistantRow({ backend, route, block, offer, picking = nu
const routeUnknown = route?.kind === 'pending' || route?.kind === 'failed';
const note = block === 'gateway_off' ? t('onboarding.setup.noteGatewayOff', { name: label })
: block === 'hub_disabled' ? t('onboarding.setup.noteHubDisabled', { name: label })
+ : block === 'engine_unsupported' ? t('onboarding.setup.noteGatewayUnsupported', { name: label })
: unset ? t('onboarding.setup.noteBuiltinUnset')
: route?.kind === 'route' && !routeModel ? t('onboarding.setup.noteNoModels')
: t('onboarding.setup.noteEnabled');
diff --git a/ui/src/components/settings/SettingsBackendPage.tsx b/ui/src/components/settings/SettingsBackendPage.tsx
index 5e769e0b8..96d5c7ce4 100644
--- a/ui/src/components/settings/SettingsBackendPage.tsx
+++ b/ui/src/components/settings/SettingsBackendPage.tsx
@@ -60,7 +60,9 @@ export function SettingsBackendPage({ backend }: { backend: AgentBackendId }) {
<p className="text-[12px] text-gold-ink">
{supplyBlock === 'hub_disabled'
? t('settings.backends.hubDisabledNotice', { name: meta.label })
- : t('settings.backends.gatewayOff', { name: meta.label })}
+ : supplyBlock === 'engine_unsupported'
+ ? t('settings.backends.gatewayUnsupported', { name: meta.label })
+ : t('settings.backends.gatewayOff', { name: meta.label })}
</p>
{supplyBlock === 'gateway_off' && <Link
to="/settings/models"
@@ -72,7 +74,8 @@ export function SettingsBackendPage({ backend }: { backend: AgentBackendId }) {
</div>
)}
/>
- {modelHubEnabled === true && <BackendSupplyModeCard backend={backend} />}
+ {/* Where no gateway engine is built, no supply mode can make this backend run. */}
+ {modelHubEnabled === true && supplyBlock !== 'engine_unsupported' && <BackendSupplyModeCard backend={backend} />}
</div>
)}
</SettingsPageShell>
diff --git a/ui/src/components/settings/SettingsBackendsPage.test.tsx b/ui/src/components/settings/SettingsBackendsPage.test.tsx
index 31668015f..0f0c2d72a 100644
--- a/ui/src/components/settings/SettingsBackendsPage.test.tsx
+++ b/ui/src/components/settings/SettingsBackendsPage.test.tsx
@@ -21,6 +21,7 @@ const mocks = vi.hoisted(() => ({
restartBackend: vi.fn(),
},
showToast: vi.fn(),
+ modelHubEnabled: false,
}));
vi.mock('@/context/ApiContext', () => ({ useApi: () => mocks.api }));
@@ -28,13 +29,14 @@ vi.mock('@/context/ToastContext', () => ({ useToast: () => ({ showToast: mocks.s
vi.mock('./SettingsPageShell', () => ({
SettingsPageShell: ({ children, title }: { children: ReactNode; title: string }) => <main><h1>{title}</h1>{children}</main>,
}));
-vi.mock('./models/useModelHubCapability', () => ({ useModelHubCapability: () => false }));
+vi.mock('./models/useModelHubCapability', () => ({ useModelHubCapability: () => mocks.modelHubEnabled }));
vi.mock('./BackendLifecycleChip', () => ({
BackendLifecycleChip: ({ name }: { name: string }) => <span data-testid={`lifecycle-${name}`} />,
}));
beforeEach(async () => {
vi.clearAllMocks();
+ mocks.modelHubEnabled = false;
await i18n.changeLanguage('en');
mocks.api.getConfig.mockResolvedValue({ agents: {} });
mocks.api.detectCli.mockImplementation(async (binary: string) => ({ found: true, path: binary }));
@@ -160,4 +162,26 @@ describe('backend attention', () => {
expect(await screen.findByText('Model Hub is disabled on this instance, so Vibey has no model to run on.')).toBeTruthy();
expect(screen.queryByRole('link', { name: 'Open Models page' })).toBeNull();
});
+
+ // No gateway engine is built for this platform: the backend stays listed and enabled,
+ // its status names that, and there is nothing on the Models page to turn on.
+ it('says this platform has no gateway component, on the row and on the backend page', async () => {
+ // The Model Hub itself stays enabled on such a host; only its engine is missing.
+ mocks.modelHubEnabled = true;
+ vi.spyOn(modelsApi, 'listAgents').mockResolvedValue(withoutModel);
+ mocks.api.getConfig.mockResolvedValue({
+ agents: {}, capabilities: { model_hub: { enabled: true } }, agent_supply_blocks: { vibey: 'engine_unsupported' },
+ });
+ renderBackends();
+
+ const vibey = await rowOf('Vibey');
+ expect(await within(vibey).findByText('No gateway component for this platform yet')).toBeTruthy();
+ expect(within(vibey).queryByText('Gateway off')).toBeNull();
+ expect(within(vibey).queryByText('No model selected')).toBeNull();
+ expect(within(await rowOf('Claude Code')).queryByText('No gateway component for this platform yet')).toBeNull();
+
+ fireEvent.click(within(vibey).getByRole('link', { name: 'Configure' }));
+ expect(await screen.findByText('No gateway component for this platform yet, so Vibey has no model to run on.')).toBeTruthy();
+ expect(screen.queryByRole('link', { name: 'Open Models page' })).toBeNull();
+ });
});
diff --git a/ui/src/components/settings/SettingsBackendsPage.tsx b/ui/src/components/settings/SettingsBackendsPage.tsx
index ffd2ff618..f381fc35a 100644
--- a/ui/src/components/settings/SettingsBackendsPage.tsx
+++ b/ui/src/components/settings/SettingsBackendsPage.tsx
@@ -195,7 +195,9 @@ export const SettingsBackendsPage: React.FC = () => {
<Badge variant="warning">
{supplyBlocks.get(meta.id) === 'hub_disabled'
? t('settings.backends.hubDisabled')
- : t('settings.models.shell.stopped')}
+ : supplyBlocks.get(meta.id) === 'engine_unsupported'
+ ? t('settings.models.shell.unsupported')
+ : t('settings.models.shell.stopped')}
</Badge>
) : agent.enabled && withoutModel.has(meta.id) && (
<Badge variant="warning">{t('settings.models.gateway.agentIssues.modelMissing')}</Badge>
diff --git a/ui/src/components/settings/models/featureFlags.ts b/ui/src/components/settings/models/featureFlags.ts
index be0b7c678..2dc4442af 100644
--- a/ui/src/components/settings/models/featureFlags.ts
+++ b/ui/src/components/settings/models/featureFlags.ts
@@ -13,7 +13,8 @@ export const modelHubEnabledFromConfig = (config: unknown): boolean => {
);
};
-export type HubSupplyBlock = 'hub_disabled' | 'gateway_off';
+const HUB_SUPPLY_BLOCKS = ['hub_disabled', 'engine_unsupported', 'gateway_off'] as const;
+export type HubSupplyBlock = typeof HUB_SUPPLY_BLOCKS[number];
// Backends left with no model to run on by the Model Hub's state, and why. The
// server decides (``hub_supply_block``) and GET /api/config carries the answer.
@@ -22,6 +23,6 @@ export const hubSupplyBlocks = (config: unknown): ReadonlyMap<string, HubSupplyB
if (!blocks || typeof blocks !== 'object') return new Map();
return new Map(
Object.entries(blocks).filter((entry): entry is [string, HubSupplyBlock] =>
- entry[1] === 'hub_disabled' || entry[1] === 'gateway_off'),
+ (HUB_SUPPLY_BLOCKS as readonly unknown[]).includes(entry[1])),
);
};
diff --git a/ui/src/i18n/en.json b/ui/src/i18n/en.json
index b95a215a5..f21a80038 100644
--- a/ui/src/i18n/en.json
+++ b/ui/src/i18n/en.json
@@ -626,6 +626,7 @@
"gatewayOff": "The model gateway is off, so {{name}} has no model to run on.",
"hubDisabled": "Model Hub disabled",
"hubDisabledNotice": "Model Hub is disabled on this instance, so {{name}} has no model to run on.",
+ "gatewayUnsupported": "No gateway component for this platform yet, so {{name}} has no model to run on.",
"builtinBadge": "Built-in",
"openProviderPage": "Configure provider",
"providerOauth": "OAuth (default)",
@@ -5368,6 +5369,7 @@
"noteBuiltinUnset": "No model yet. Pick one to start.",
"noteGatewayOff": "Model gateway is off. {{name}} needs it to run.",
"noteHubDisabled": "Model Hub is disabled on this instance. {{name}} needs it to run.",
+ "noteGatewayUnsupported": "No gateway component for this platform yet. {{name}} needs it to run.",
"pickModelNamed": "Pick a model for {{name}}",
"allModels": "All models…",
"modelsUnreadable": "Could not read the models.",
diff --git a/ui/src/i18n/zh.json b/ui/src/i18n/zh.json
index 2f83fa405..aa6a948ff 100644
--- a/ui/src/i18n/zh.json
+++ b/ui/src/i18n/zh.json
@@ -626,6 +626,7 @@
"gatewayOff": "模型网关已关闭,{{name}} 没有可用的模型。",
"hubDisabled": "模型网关已停用",
"hubDisabledNotice": "本实例已停用模型网关,{{name}} 没有可用的模型。",
+ "gatewayUnsupported": "当前平台不支持模型网关,{{name}} 没有可用的模型。",
"builtinBadge": "内置",
"openProviderPage": "配置 Provider",
"providerOauth": "OAuth(默认)",
@@ -5368,6 +5369,7 @@
"noteBuiltinUnset": "还没有模型,选一个即可开始",
"noteGatewayOff": "模型网关已关闭,{{name}} 需要它才能运行",
"noteHubDisabled": "本实例已停用模型网关,{{name}} 需要它才能运行",
+ "noteGatewayUnsupported": "当前平台不支持模型网关,{{name}} 需要它才能运行",
"pickModelNamed": "为 {{name}} 选择模型",
"allModels": "全部模型…",
"modelsUnreadable": "未能读取模型",
diff --git a/vibe/i18n/en.json b/vibe/i18n/en.json
index a3f565f78..83e3fab66 100644
--- a/vibe/i18n/en.json
+++ b/vibe/i18n/en.json
@@ -7,6 +7,7 @@
"errors": {
"agentModelRequired": "Select a model for the {backend} Agent or Session in Avibe before running it.",
"modelHubDisabled": "Model Hub is disabled on this instance, so {backend} has no model to run on.",
+ "modelGatewayUnsupported": "No gateway component for this platform yet, so {backend} has no model to run on.",
"modelGatewayOff": "The model gateway is off, so {backend} has no model to run on. Turn it on in Models, then choose a model.",
"opencodeModelProviderRequired": "Select an OpenCode model with a provider in Avibe, or configure its Model Hub route."
},
diff --git a/vibe/i18n/zh.json b/vibe/i18n/zh.json
index 9084eb0f0..acddfe0b1 100644
--- a/vibe/i18n/zh.json
+++ b/vibe/i18n/zh.json
@@ -7,6 +7,7 @@
"errors": {
"agentModelRequired": "请先在 Avibe 中为 {backend} Agent 或会话选择模型。",
"modelHubDisabled": "本实例已停用模型网关,{backend} 没有可用的模型。",
+ "modelGatewayUnsupported": "当前平台不支持模型网关,{backend} 没有可用的模型。",
"modelGatewayOff": "模型网关已关闭,{backend} 没有可用的模型。请在「模型」页开启网关并选择模型。",
"opencodeModelProviderRequired": "请在 Avibe 中选择带供应商的 OpenCode 模型,或配置它的 Model Hub 路由。"
},