311 tools start failing but the city still wants resident trust. What uptime and transparency practice fits?

When 311 tools start failing, resident trust depends on honest uptime communication—not silent degradation. Showing status pages for agent degraded mode when tools fail preserves trust by stating what still works (for example, FAQ answers) and what does not (live case lookups, ticket updates). Degraded-mode status pages align municipal transparency with operational reality: residents and clerks see the outage class, expected impact, and recovery posture instead of discovering failures through wrong answers or unexplained errors.

Hiding all failures and pretending the agent is fully healthy forever fails by concept because false confidence converts an infrastructure outage into silent mis-service and destroys credibility when residents compare answers to known city facts. Taking the entire city portal offline with no status message fails because a hard blackout without communication maximizes confusion and support load; transparency still requires messaging even when functionality is reduced. Replying with random confident answers when tools are down fails because hallucinated or fabricated tool results are worse than admitting degraded mode—they create incorrect civic guidance during an already broken dependency path.

Exam caveat: NCP-AAI reliability and transparency questions favor explicit degraded-mode communication over denial, unexplained outages, or confident fabrication. Status pages are the operational artifact that couples uptime practice with resident trust.

Operational check: wire tool-health probes to a public status page that switches the 311 agent into labeled degraded mode; verify copy lists unavailable capabilities; confirm the agent refuses or soft-fails tool-dependent actions instead of inventing results; exercise a tabletop where tools fail and staff confirm the status page updates within the city’s SLA.