Repository navigation
Upgrading
sync migrates your config automatically. This page tells you what changed so nothing surprises you.
agent-bridge sync always fetched sources first, so update was redundant. It is now a hidden alias that runs sync.
- Git hooks installed by older versions ran
agent-bridge update && agent-bridge sync. They still work. The firstsyncrewrites hooks carrying the Agent Bridge marker to runsynconly. - Update any scripts or
postinstalllines to callsyncalone.
Before:
domains: [backend, shared]
sources:
- name: hub
source: https://github.com/org/hub.gitAfter (written for you on the first sync):
sources:
- name: hub
source: https://github.com/org/hub.git
domains:
- name: backend
- name: sharedBehavior is identical: every listed domain, everything inside it. The old top-level form keeps loading as a fallback, so an un-migrated file never breaks.
domains:
- name: shared
include:
- skills/code-review
- AGENTS.mdSee Configuration. Manifests and sync behavior are unchanged: anything you deselect is removed on the next sync, like a feature deleted upstream.
init now asks for tools and sources first, fetches the sources, and shows the actual domains in each one as a grouped checklist. The default domain list backend,frontend,shared is gone; non-interactive init without --domains takes every domain it finds.
A config with a newer version than the installed package is left alone. The 0.14 per-source domains field is rejected by 0.13 and earlier, so pin the package version across the team.