Skip to content

Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS

test([issue-12769]): pin the prompt-not-submitted verdict as a retryable startup failure

[._.]/ Adam Eivyf5a9c6805cfa
diff --git a/server/services/agentErrorAnalysis.test.js b/server/services/agentErrorAnalysis.test.js
index 9d254cd561..7204a06593 100644
--- a/server/services/agentErrorAnalysis.test.js
+++ b/server/services/agentErrorAnalysis.test.js
@@ -1367,9 +1367,11 @@ describe('analyzeAgentFailure — TUI startup gates are not lint errors', () =>
expect(analysis.suggestedFix).not.toMatch(/idle reaper/i);
});
- it('registers both spawner-resolved startup verdicts', () => {
+ it('registers the spawner-resolved startup verdicts', () => {
expect(COMPLETION_REASON_ANALYSES['paste-not-rendered'].category).toBe('startup-failure');
expect(COMPLETION_REASON_ANALYSES['tui-not-ready'].category).toBe('startup-failure');
+ // Non-actionable, so the task retries (#12769).
+ expect(COMPLETION_REASON_ANALYSES['prompt-not-submitted']).toMatchObject({ category: 'startup-failure', actionable: false });
});
// Defense in depth for the case with no completion reason to lean on: the