Skip to content

Evenfire is an intelligence layer platform for running LLM agents that take real actions, on your infrastructure

TypeScript

test(control-api): allow Codex stale exclusion in model materializer

Alfredo Lopezeff070c56b33
diff --git a/control-api/test/services.llmAllowedModels.test.ts b/control-api/test/services.llmAllowedModels.test.ts
index cbe8d688b..c18486f17 100644
--- a/control-api/test/services.llmAllowedModels.test.ts
+++ b/control-api/test/services.llmAllowedModels.test.ts
@@ -341,8 +341,10 @@ describe('llmAllowedModels service', () => {
})
it('materializes an enabled row even when it is stale (R3.7: stale never de-serves)', async () => {
- // The materializer query filters `WHERE enabled` and never references
- // `stale`, so an enabled model discovery flagged stale still reaches the CM.
+ // R3.7: enabled non-Codex rows still reach the ConfigMap even if discovery
+ // flagged them stale. Codex subscription is the documented exception:
+ // a stale row must not be served because the broker catalog is the only
+ // live model source.
const query = vi.fn().mockResolvedValue({
rows: [
{
@@ -358,7 +360,8 @@ describe('llmAllowedModels service', () => {
const grouped = await listEnabledGroupedByProvider(fakeDb(query))
const [sql] = query.mock.calls[0]
expect(String(sql)).toMatch(/WHERE enabled/)
- expect(String(sql)).not.toMatch(/stale/)
+ expect(String(sql)).toContain("NOT (provider = 'codex-subscription' AND stale)")
+ expect(String(sql)).not.toMatch(/AND NOT stale\b/)
expect(grouped.openai).toEqual([{ model: 'gpt-legacy' }])
})
})