fix: post-finale completeness sweep — routing surface, provider config, budgets, compose env, interactive exemption (#661)

This commit is contained in:
Renzo F
2026-07-23 09:41:27 +02:00
committed by GitHub
parent 21d6730400
commit d4b7e1e7b8
45 changed files with 2064 additions and 256 deletions
@@ -50,6 +50,11 @@ const {
provider_type: "openai",
display_name: "GPT-5.3 Codex",
},
{
model_name: "gemini-2.5-pro",
provider_type: "gemini",
display_name: "Gemini 2.5 Pro",
},
]),
getOllamaKey: vi.fn(async () => ({ has_key: false, enabled: true })),
setOllamaKey: vi.fn(async () => ({ has_key: true, enabled: true })),
@@ -387,6 +392,13 @@ function withQueryClient(ui: ReactNode) {
return <QueryClientProvider client={client}>{ui}</QueryClientProvider>;
}
// Finds a per-agent Mix row's container div by its agent-id text. `.closest`
// on a non-tag-name CSS selector types as `Element | null`, not `HTMLElement`
// — cast once here rather than at every call site.
function mixRowFor(agentId: string): HTMLElement {
return screen.getByText(agentId).closest("div.grid") as HTMLElement;
}
describe("AIRoutingCard", () => {
beforeEach(() => {
catalog.mockClear();
@@ -761,6 +773,102 @@ describe("AIRoutingCard", () => {
});
});
// -------------------------------------------------------------------------
// Codex and Gemini mode buttons + Mix picker visibility (the headline gap:
// both were built but unreachable from the panel — no apply-mode card, no
// Mix group).
// -------------------------------------------------------------------------
describe("Codex and Gemini mode buttons", () => {
it("renders the Codex button and applies mode='codex' on confirm", async () => {
const confirmSpy = vi.spyOn(window, "confirm").mockReturnValue(true);
render(withQueryClient(<AIRoutingCard />));
await screen.findByText("Grok (xAI) API key");
fireEvent.click(screen.getByText("Codex"));
await waitFor(() =>
expect(applyMode).toHaveBeenCalledWith({ mode: "codex" }),
);
confirmSpy.mockRestore();
});
it("renders the Gemini button and applies mode='gemini' on confirm", async () => {
const confirmSpy = vi.spyOn(window, "confirm").mockReturnValue(true);
render(withQueryClient(<AIRoutingCard />));
await screen.findByText("Grok (xAI) API key");
fireEvent.click(screen.getByText("Gemini"));
await waitFor(() =>
expect(applyMode).toHaveBeenCalledWith({ mode: "gemini" }),
);
confirmSpy.mockRestore();
});
it("neither button is gated on a key (no key card exists for either provider)", async () => {
render(withQueryClient(<AIRoutingCard />));
await screen.findByText("Grok (xAI) API key");
expect(screen.getByText("Codex").closest("button")).not.toBeDisabled();
expect(screen.getByText("Gemini").closest("button")).not.toBeDisabled();
});
});
describe("Mix picker Codex/Gemini group visibility", () => {
it("shows Codex and Gemini provider groups for a delivery role's per-agent select", async () => {
render(withQueryClient(<AIRoutingCard />));
await screen.findByText("Per-agent override (mix mode)");
const beDevRow = mixRowFor("be-dev-1");
// The catalog query resolves asynchronously — the per-agent groups are
// absent on the first render pass, so wait for them (findByText) rather
// than asserting synchronously.
expect(
await within(beDevRow).findByText("Codex (OpenAI)"),
).toBeInTheDocument();
expect(
within(beDevRow).getByText("Gemini (Google)"),
).toBeInTheDocument();
});
it("excludes Codex and Gemini from the Intake/Secretary/PR Review group, with an inline note", async () => {
render(withQueryClient(<AIRoutingCard />));
await screen.findByText("Per-agent override (mix mode)");
expect(
screen.getByText(/Codex and Gemini are delivery-roles-only/i),
).toBeInTheDocument();
// Wait for the catalog query to resolve (an unrelated row's groups)
// before asserting absence on this group's rows below.
await within(mixRowFor("be-dev-1")).findByText("Codex (OpenAI)");
const secretaryRow = mixRowFor("secretary-1");
expect(
within(secretaryRow).queryByText("Codex (OpenAI)"),
).not.toBeInTheDocument();
expect(
within(secretaryRow).queryByText("Gemini (Google)"),
).not.toBeInTheDocument();
const intakeRow = mixRowFor("intake-1");
expect(
within(intakeRow).queryByText("Codex (OpenAI)"),
).not.toBeInTheDocument();
expect(
within(intakeRow).queryByText("Gemini (Google)"),
).not.toBeInTheDocument();
// The root PR reviewer shares the same group/note, even though it is
// technically one-shot-capable — the panel restricts the whole group.
const prReviewerRow = mixRowFor("pr-reviewer-1");
expect(
within(prReviewerRow).queryByText("Codex (OpenAI)"),
).not.toBeInTheDocument();
});
});
// -------------------------------------------------------------------------
// Mode switches preserve complexity overrides (2026-07-17-style incident:
// these same buttons once wiped AGENT_SLUG pins) — the confirm text says so
+271 -175
View File
@@ -40,8 +40,10 @@ import {
import { Separator } from "@/components/ui/separator";
import {
AlertTriangle,
Bot,
Cpu,
Gauge,
Gem,
Key,
KeyRound,
Server,
@@ -115,6 +117,12 @@ const AGENT_GROUP_DEFS: {
},
];
// Codex/Gemini are V1 delivery-roles-only — no interactive Intake/Secretary
// support (see roboco.llm.providers.codex / .gemini). This group's per-agent
// picker excludes both providers below instead of offering a route that
// would silently misroute the persistent Intake/Secretary session at spawn.
const INTERACTIVE_ONLY_GROUP_TITLE = "Intake / Secretary / PR Review";
// Stable within-group ordering (PM/lead first, devs, QA, doc, reviewer last)
// so the picker doesn't churn alphabetically as the live roster loads —
// ties (e.g. dev-1/dev-2) break on slug, which already sorts correctly.
@@ -280,6 +288,10 @@ export function AIRoutingCard() {
(c: { provider_type: ModelProvider }) =>
c.provider_type === ModelProvider.OPENAI,
);
const catalogGeminiOnly = catalog.filter(
(c: { provider_type: ModelProvider }) =>
c.provider_type === ModelProvider.GEMINI,
);
const catalogAnthropicOnly = catalog.filter(
(c: { provider_type: ModelProvider }) =>
c.provider_type === ModelProvider.ANTHROPIC,
@@ -326,6 +338,46 @@ export function AIRoutingCard() {
}
};
const flipToCodex = async () => {
if (
!confirm(
"Switch every agent to Codex? Per-agent pins and complexity " +
"overrides are kept; other role/global assignments are replaced. " +
"Intake and Secretary stay on Anthropic (Codex has no interactive " +
"chat support).",
)
)
return;
try {
await applyMode.mutateAsync({ mode: "codex" });
toast.success(
"Role/global routing now on Codex — pins/overrides kept, Intake & Secretary stay on Anthropic",
);
} catch (e) {
toast.error("Switch failed: " + errMsg(e));
}
};
const flipToGemini = async () => {
if (
!confirm(
"Switch every agent to Gemini? Per-agent pins and complexity " +
"overrides are kept; other role/global assignments are replaced. " +
"Intake and Secretary stay on Anthropic (Gemini has no interactive " +
"chat support).",
)
)
return;
try {
await applyMode.mutateAsync({ mode: "gemini" });
toast.success(
"Role/global routing now on Gemini — pins/overrides kept, Intake & Secretary stay on Anthropic",
);
} catch (e) {
toast.error("Switch failed: " + errMsg(e));
}
};
const flipToOllama = async () => {
if (!hasOllamaKey) {
toast.error("Save an Ollama API key first");
@@ -551,6 +603,126 @@ export function AIRoutingCard() {
}
};
// The full per-agent model-picker option list, shared by every group's
// Select — factored out so the Codex/Gemini exclusion for the interactive
// group (`restrictInteractiveOnly`) doesn't require duplicating the whole
// catalog-grouped SelectContent tree.
const renderMixSelectOptions = (restrictInteractiveOnly: boolean) => (
<>
<SelectItem value="__clear__">(inherit global)</SelectItem>
{/* Anthropic models */}
{catalogAnthropicOnly.length > 0 && (
<SelectGroup>
<SelectLabel>
<ProviderBadge variant="anthropic" />
Anthropic
</SelectLabel>
{catalogAnthropicOnly.map(
(c: { model_name: string; display_name: string }) => (
<SelectItem key={c.model_name} value={c.model_name}>
{c.display_name}
</SelectItem>
),
)}
</SelectGroup>
)}
{/* Grok (xAI) models */}
{catalogGrokOnly.length > 0 && (
<SelectGroup>
<SelectLabel>
<ProviderBadge variant="grok" />
Grok (xAI)
</SelectLabel>
{catalogGrokOnly.map(
(c: { model_name: string; display_name: string }) => (
<SelectItem key={c.model_name} value={c.model_name}>
{c.display_name}
</SelectItem>
),
)}
</SelectGroup>
)}
{/* Codex (OpenAI) models — excluded for the interactive-only group */}
{!restrictInteractiveOnly && catalogOpenaiOnly.length > 0 && (
<SelectGroup>
<SelectLabel>
<ProviderBadge variant="openai" />
Codex (OpenAI)
</SelectLabel>
{catalogOpenaiOnly.map(
(c: { model_name: string; display_name: string }) => (
<SelectItem key={c.model_name} value={c.model_name}>
{c.display_name}
</SelectItem>
),
)}
</SelectGroup>
)}
{/* Gemini (Google) models — excluded for the interactive-only group */}
{!restrictInteractiveOnly && catalogGeminiOnly.length > 0 && (
<SelectGroup>
<SelectLabel>
<ProviderBadge variant="gemini" />
Gemini (Google)
</SelectLabel>
{catalogGeminiOnly.map(
(c: { model_name: string; display_name: string }) => (
<SelectItem key={c.model_name} value={c.model_name}>
{c.display_name}
</SelectItem>
),
)}
</SelectGroup>
)}
{/* Ollama Cloud models */}
{catalogOllamaOnly.length > 0 && (
<SelectGroup>
<SelectLabel>
<ProviderBadge variant="ollama" />
Ollama Cloud
</SelectLabel>
{catalogOllamaOnly.map(
(c: { model_name: string; display_name: string }) => (
<SelectItem key={c.model_name} value={c.model_name}>
{c.display_name}
</SelectItem>
),
)}
</SelectGroup>
)}
{/* Self-Hosted models */}
{selfHostedModels.length > 0 && (
<SelectGroup>
<SelectLabel>
<ProviderBadge variant="self-hosted" />
Self-Hosted
</SelectLabel>
{selfHostedModels.map((m: SelfHostedModel) => (
<SelectItem key={m.model_name} value={m.model_name}>
{m.display_name}
</SelectItem>
))}
</SelectGroup>
)}
{/* Fallback: un-grouped catalog when no grouping is possible */}
{catalogAnthropicOnly.length === 0 &&
catalogOllamaOnly.length === 0 &&
selfHostedModels.length === 0 &&
catalogForMix.map((c: { model_name: string; display_name: string }) => (
<SelectItem key={c.model_name} value={c.model_name}>
{c.display_name} {c.model_name}
</SelectItem>
))}
</>
);
return (
<Card>
<CardHeader>
@@ -560,8 +732,10 @@ export function AIRoutingCard() {
<CardDescription>
Decide which model backs each agent. Anthropic uses the mounted
<code className="px-1"> ~/.claude </code> auth; Grok (xAI) and Ollama
Cloud use the API keys you save below; Self-Hosted connects to any
OpenAI-compatible endpoint you run locally.
Cloud use the API keys you save below; Codex and Gemini authenticate
via their own mounted CLI subscriptions (no key needed) V1:
delivery roles only, not Intake/Secretary; Self-Hosted connects to
any OpenAI-compatible endpoint you run locally.
</CardDescription>
</CardHeader>
<CardContent className="space-y-6">
@@ -700,10 +874,10 @@ export function AIRoutingCard() {
{/* -------- Mode toggle -------- */}
<section className="space-y-3">
<HelpTip label="Anthropic / Grok / Ollama / Self-Hosted replace role/global routing with that provider; per-agent pins in the table below survive the switch. Mix keeps whatever's picked in the table.">
<HelpTip label="Anthropic / Grok / Codex / Gemini / Ollama / Self-Hosted replace role/global routing with that provider; per-agent pins in the table below survive the switch. Mix keeps whatever's picked in the table.">
<Label className="text-sm font-medium">Routing mode</Label>
</HelpTip>
<div className="grid grid-cols-2 md:grid-cols-3 lg:grid-cols-6 gap-2">
<div className="grid grid-cols-2 md:grid-cols-4 lg:grid-cols-8 gap-2">
<ModeButton
icon={<ShieldCheck className="h-4 w-4" />}
label="Anthropic"
@@ -724,6 +898,24 @@ export function AIRoutingCard() {
onClick={flipToGrok}
disabled={applyMode.isPending || !hasGrokKey}
/>
<ModeButton
icon={<Bot className="h-4 w-4" />}
label="Codex"
description="Every agent uses Codex (gpt-5.3-codex)."
active={currentMode === "codex"}
onClick={flipToCodex}
disabled={applyMode.isPending}
labelHint="Codex authenticates via a mounted ~/.codex subscription (ChatGPT, no API key) — always available once the CLI is logged in on the host. V1: delivery roles only, not offered for Intake/Secretary."
/>
<ModeButton
icon={<Gem className="h-4 w-4" />}
label="Gemini"
description="Every agent uses Gemini (gemini-2.5-pro)."
active={currentMode === "gemini"}
onClick={flipToGemini}
disabled={applyMode.isPending}
labelHint="Gemini authenticates via a mounted ~/.gemini OAuth login (no API key) — always available once the CLI is logged in on the host. V1: delivery roles only, not offered for Intake/Secretary."
/>
<ModeButton
icon={<Sparkles className="h-4 w-4" />}
label="Ollama"
@@ -784,6 +976,22 @@ export function AIRoutingCard() {
per-agent cost cap all apply.
</p>
) : null}
{currentMode === "codex" || currentMode === "mix" ? (
<p className="text-xs text-muted-foreground">
Codex agents run on OpenAI&apos;s official Codex CLI (ChatGPT
subscription, mounted ~/.codex); the same command /
secret-exfiltration guard, prompt-injection guard, and per-agent
cost cap apply. V1: delivery roles only not available for
Intake/Secretary.
</p>
) : null}
{currentMode === "gemini" || currentMode === "mix" ? (
<p className="text-xs text-muted-foreground">
Gemini agents run on Google&apos;s official gemini CLI (OAuth
login, mounted ~/.gemini); the same guards apply. V1: delivery
roles only not available for Intake/Secretary.
</p>
) : null}
</section>
{/* -------- Self-Hosted model picker (when self_hosted mode active) -------- */}
@@ -953,178 +1161,58 @@ export function AIRoutingCard() {
</div>
) : (
<div className="divide-y rounded-md border">
{agentGroups.map((group) => (
<div key={group.title} className="p-4">
<HelpTip label={group.titleHint}>
<h4 className="mb-2 w-fit text-xs font-semibold text-muted-foreground uppercase tracking-wider">
{group.title}
</h4>
</HelpTip>
<div className="grid grid-cols-1 gap-x-8 gap-y-3 sm:grid-cols-2">
{group.agents.map((a) => (
<div
key={a.id}
className="grid grid-cols-[1fr_170px] items-center gap-4 rounded-md border px-3 py-2.5"
>
<div className="min-w-0">
<div className="truncate font-mono text-xs">
{a.id}
</div>
<div className="truncate text-[11px] text-muted-foreground">
{a.name}
</div>
</div>
<Select
value={mixMap[a.id] ?? ""}
onValueChange={(v: string) =>
setMixMap((prev) => ({
...prev,
[a.id]: v === "__clear__" ? "" : v,
}))
}
{agentGroups.map((group) => {
const restrictInteractiveOnly =
group.title === INTERACTIVE_ONLY_GROUP_TITLE;
return (
<div key={group.title} className="p-4">
<HelpTip label={group.titleHint}>
<h4 className="mb-2 w-fit text-xs font-semibold text-muted-foreground uppercase tracking-wider">
{group.title}
</h4>
</HelpTip>
{restrictInteractiveOnly ? (
<p className="mb-2 text-[11px] text-muted-foreground">
Codex and Gemini are delivery-roles-only (V1) not
offered here (no interactive Intake/Secretary support).
</p>
) : null}
<div className="grid grid-cols-1 gap-x-8 gap-y-3 sm:grid-cols-2">
{group.agents.map((a) => (
<div
key={a.id}
className="grid grid-cols-[1fr_170px] items-center gap-4 rounded-md border px-3 py-2.5"
>
<SelectTrigger size="sm" className="w-full text-xs">
<SelectValue placeholder="(inherit)" />
</SelectTrigger>
<SelectContent>
<SelectItem value="__clear__">
(inherit global)
</SelectItem>
{/* Anthropic models */}
{catalogAnthropicOnly.length > 0 && (
<SelectGroup>
<SelectLabel>
<ProviderBadge variant="anthropic" />
Anthropic
</SelectLabel>
{catalogAnthropicOnly.map(
(c: {
model_name: string;
display_name: string;
}) => (
<SelectItem
key={c.model_name}
value={c.model_name}
>
{c.display_name}
</SelectItem>
),
)}
</SelectGroup>
)}
{/* Grok (xAI) models */}
{catalogGrokOnly.length > 0 && (
<SelectGroup>
<SelectLabel>
<ProviderBadge variant="grok" />
Grok (xAI)
</SelectLabel>
{catalogGrokOnly.map(
(c: {
model_name: string;
display_name: string;
}) => (
<SelectItem
key={c.model_name}
value={c.model_name}
>
{c.display_name}
</SelectItem>
),
)}
</SelectGroup>
)}
{/* Codex (OpenAI) models */}
{catalogOpenaiOnly.length > 0 && (
<SelectGroup>
<SelectLabel>
<ProviderBadge variant="openai" />
Codex (OpenAI)
</SelectLabel>
{catalogOpenaiOnly.map(
(c: {
model_name: string;
display_name: string;
}) => (
<SelectItem
key={c.model_name}
value={c.model_name}
>
{c.display_name}
</SelectItem>
),
)}
</SelectGroup>
)}
{/* Ollama Cloud models */}
{catalogOllamaOnly.length > 0 && (
<SelectGroup>
<SelectLabel>
<ProviderBadge variant="ollama" />
Ollama Cloud
</SelectLabel>
{catalogOllamaOnly.map(
(c: {
model_name: string;
display_name: string;
}) => (
<SelectItem
key={c.model_name}
value={c.model_name}
>
{c.display_name}
</SelectItem>
),
)}
</SelectGroup>
)}
{/* Self-Hosted models */}
{selfHostedModels.length > 0 && (
<SelectGroup>
<SelectLabel>
<ProviderBadge variant="self-hosted" />
Self-Hosted
</SelectLabel>
{selfHostedModels.map((m: SelfHostedModel) => (
<SelectItem
key={m.model_name}
value={m.model_name}
>
{m.display_name}
</SelectItem>
))}
</SelectGroup>
)}
{/* Fallback: un-grouped catalog when no grouping is possible */}
{catalogAnthropicOnly.length === 0 &&
catalogOllamaOnly.length === 0 &&
selfHostedModels.length === 0 &&
catalogForMix.map(
(c: {
model_name: string;
display_name: string;
}) => (
<SelectItem
key={c.model_name}
value={c.model_name}
>
{c.display_name} {c.model_name}
</SelectItem>
),
)}
</SelectContent>
</Select>
</div>
))}
<div className="min-w-0">
<div className="truncate font-mono text-xs">
{a.id}
</div>
<div className="truncate text-[11px] text-muted-foreground">
{a.name}
</div>
</div>
<Select
value={mixMap[a.id] ?? ""}
onValueChange={(v: string) =>
setMixMap((prev) => ({
...prev,
[a.id]: v === "__clear__" ? "" : v,
}))
}
>
<SelectTrigger size="sm" className="w-full text-xs">
<SelectValue placeholder="(inherit)" />
</SelectTrigger>
<SelectContent>
{renderMixSelectOptions(restrictInteractiveOnly)}
</SelectContent>
</Select>
</div>
))}
</div>
</div>
</div>
))}
);
})}
</div>
)}
{catalogOllamaOnly.length === 0 ? (
@@ -1274,7 +1362,13 @@ function errMsg(e: unknown): string {
function ProviderBadge({
variant,
}: {
variant: "anthropic" | "grok" | "openai" | "ollama" | "self-hosted";
variant:
| "anthropic"
| "grok"
| "openai"
| "gemini"
| "ollama"
| "self-hosted";
}) {
const styles: Record<string, string> = {
anthropic: "bg-blue-500/20 text-blue-700 dark:text-blue-400",
@@ -1282,6 +1376,7 @@ function ProviderBadge({
"self-hosted": "bg-purple-500/20 text-purple-700 dark:text-purple-400",
grok: "bg-teal-500/20 text-teal-700 dark:text-teal-400",
openai: "bg-emerald-500/20 text-emerald-700 dark:text-emerald-400",
gemini: "bg-sky-500/20 text-sky-700 dark:text-sky-400",
};
const labels: Record<string, string> = {
anthropic: "A",
@@ -1289,6 +1384,7 @@ function ProviderBadge({
"self-hosted": "S",
grok: "G",
openai: "C",
gemini: "Ge",
};
return (
<span