redshift 7d95a12f84 fix(onboard): wait for models and warn instead of aborting ...

`startDaemon` only waits for `/health`, so onboarding reached the
integration step before models were loaded. On a cold start `/models`
blocks while providers bootstrap, and can still come back empty when
relays are slow. Since the installers now throw, that ended onboarding
with a failure after the daemon and API key were already created.

- Wait for a non-empty `/models` before installing OpenCode, OpenClaw and
  Pi (the integrations whose config embeds the model list), retrying for
  up to 5 min with a progress line every 10s.
- Catch installer failures during onboarding and print a warning that
  points at `routstrd refresh`, so onboarding still reports completion.
  The installers keep throwing, so scheduled/bulk refresh still counts
  real failures.

2026-10-13 19:58:41 +08:00

..

daemon

chore(wallet): remove the dead legacy cocod client

2026-10-02 16:59:42 +08:00

integrations

fix(onboard): wait for models and warn instead of aborting

2026-10-13 19:58:41 +08:00

utils

fix(update): bound downloads with a two-tier deadline

2026-10-03 11:12:38 +08:00

wallet

chore(wallet): remove the dead legacy cocod client

2026-10-02 16:59:42 +08:00

daemon-command.test.ts

feat: add standalone binary releases (#89)

2026-09-08 15:23:36 +00:00

install-script.test.ts

feat(release): add install.sh for standalone binaries (#101)

2026-09-10 20:26:57 +02:00

runtime.test.ts

feat: add standalone binary releases (#89)

2026-09-08 15:23:36 +00:00

update-notice.test.ts

fix(update): configure checks and validate cached notices

2026-10-13 15:49:50 +08:00