Who
Operators hitting Hustle-Launch/mission-control-os after Google OAuth who saw AgencyGate / PortalGate / cockpit flash signed-out while Clerk was still pending, or sat on infinite Working during agency create/select.
This is the mission-control-os web product, not the Go p10k TUI and not hurley-mission-control.
What
On Hustle-Launch/mission-control-os, PR #64 merged as d09322c and closed Issues #50 and #51.
Shared auth hooks keep pending sessions from looking signed-out on AgencyGate, PortalGate, and cockpit. Agency org create/select now has a hard timeout and surfaces an error instead of infinite Working. PortalGate pending shows a setup CTA.
PR CI was green. Prod redeploy is still blocked on an invalid VERCEL_TOKEN secret (infra, separate from this code fix).
Where
Repo: https://github.com/Hustle-Launch/mission-control-os
PR: https://github.com/Hustle-Launch/mission-control-os/pull/64
Sibling names to keep straight: /blog/mission-control-go-tui-p10k-portfolio-ops (Go TUI) and /blog/hurley-mission-control-human-agent-comms (Convex comms plane).
When
September 8, 2026 evening ET: PR #64 lands (d09322c). Issues #50 and #51 CLOSED. Prod token rotation still open.
Why
Pending Clerk sessions that render as signed-out train operators to re-auth loops. Infinite Working on agency create hides real failures. The code path is fixed; the Vercel token is the remaining ship gate.
Would you rotate VERCEL_TOKEN first, or smoke AgencyGate against a preview deploy before touching prod?