.agents/skills/sync-main-to-next/SKILL.md
Handle $ARGUMENTS by loading .agents/rules/release-lanes.mdc and running only
its Sync Main To Next lane.
This is a convenience wrapper, not a second release system. release-lanes
owns the release architecture, script contract, hard stops, and handoff shape.
Default to the real sync when the user says sync-main-to-next without
arguments.
Use this sequence:
node tooling/scripts/release-branch-prs.mjs sync-main-to-next --dry-run
node tooling/scripts/release-branch-prs.mjs sync-main-to-next --push
node tooling/scripts/release-branch-prs.mjs verify-main-to-next-sync
Then watch only the workflows created by the pushed sync commit:
gh run list --branch next --limit 10 \
--json databaseId,name,workflowName,headSha,status,conclusion,displayTitle,url
gh run watch <release-run-id> --exit-status
gh run watch <ci-run-id> --exit-status
Read back:
npm view platejs dist-tags --json
npm view platejs@latest version
npm view platejs@beta version
gh release list --limit 5
gh pr list --base next --head sync/main-to-next --state open --json number,url
Close stale sync/main-to-next PRs only after the direct sync and release
verification pass.
main -> next PR.next -> main.Use the release-lanes hard stops. In practice, stop only for:
latest or betaReport in five lines or fewer:
latest and beta