fix(grok): fence project permission settings and extend qualification

This commit is contained in:
Dotta committed 2026-09-22 16:42:05 -05:00
1 parent 6adb00bcfe
commit 7d95022f34
6 files changed
+89 -13

No files matched your search

+13 -3
View File
@@ -71,10 +71,10 @@ describe("runner E2E catalog", () => {
expect(localIntegrityTasks).toHaveLength(2);
expect(openRouterBreadthTasks).toHaveLength(3);
expect(runnerSuites.map((suite) => suite.expectedMatrixSize)).toEqual([
10, 23, 38, 52, 28, 18, 6, 6, 48, 16, 10, 2,
16, 23, 38, 52, 28, 18, 6, 6, 48, 16, 10, 2,
]);
expect(validateRunnerCatalog()).toHaveLength(257);
expect(new Set(runnerMatrix.map((entry) => entry.id)).size).toBe(257);
expect(validateRunnerCatalog()).toHaveLength(263);
expect(new Set(runnerMatrix.map((entry) => entry.id)).size).toBe(263);
expect(
runnerMatrix.filter((entry) => entry.suite.id === "core-compatibility"),
).toHaveLength(48);
@@ -550,6 +550,16 @@ describe("runner E2E selectors", () => {
]);
});
it("keeps Grok artifact and stop/resume qualification explicit in both environments", () => {
const selected = selectRunnerExecutions(parseRunnerSelectors(["--suite", "grok-qualification"]));
expect(selected).toHaveLength(16);
for (const environment of ["local", "daytona"]) {
for (const task of ["build-revise", "stop-new-resume", "continuity-restart"]) {
expect(selected.some(execution => execution.id === `grok-qualification.runner-acpx-grok.${environment}.${task}`)).toBe(true);
}
}
});
it("selects a suite without exploding its environment matrix", () => {
const selected = selectRunnerExecutions(
parseRunnerSelectors(["--suite", "openrouter-model-breadth"]),
+7 -3
View File
@@ -919,12 +919,16 @@ const everydayProfiles = [
export const runnerSuites: readonly RunnerSuiteFixture[] = [
{
id: "grok-qualification", label: "Grok Build Qualification", manualOnly: true,
description: "Grok task replies, planning approval, questions and controller restart in local and Daytona environments.",
description: "Grok replies, planning approval, questions, downloadable artifacts, stop/resume and controller restart in local and Daytona environments.",
groups: ["native"],
profiles: runnerProfiles.filter(profile => profile.id === "runner-acpx-grok"),
environments: [localEnvironment, daytonaWarmEnvironment],
tasks: [...runnerTasks, ...localIntegrityTasks], expectedMatrixSize: 10,
definitionMetadata: { version: 1, binary: "1.0.13", model: "grok-4.7", scheduling: "explicit-only", repetitionsRequired: 3 },
tasks: [
...runnerTasks, ...localIntegrityTasks,
...everydayTasks.filter(task => task.id === "build-revise"),
...chatHardeningTasks.filter(task => ["stop-new-resume", "continuity-restart"].includes(task.id)),
], expectedMatrixSize: 16,
definitionMetadata: { version: 2, binary: "1.0.13", model: "grok-4.7", scheduling: "explicit-only", repetitionsRequired: 3, artifactOracle: "independent-python-contract", stopBoundary: "provider-turn-started" },
},
{
id: "continuation", label: "Task continuation",