sag-sync-projectlisted

# sag-sync-project — batch sync (manual trigger only) This skill **does not load itself** — it only runs when the user types `/sag-sync-project` or when the orchestrator explicitly assigns you the task of running a sync. This is an action on an ENTIRE SET of documents (unlike `sag-publish`, which acts on a single document you just created). ## Required process: always dry-run first ```bash sagctl sync --manifest .sag-sync.json # default = dry-run, no --yes ``` Read the result, report to the user the list of what will be published/queued/skipped. **Only run for real once the user confirms** the dry-run list is what they intend: ```bash SAGCTL_ALLOW_SYNC=1 sagctl sync --manifest .sag-sync.json --yes ``` (The `SAGCTL_ALLOW_SYNC` variable must be set explicitly — this is a signal that the `sync` operation has been deliberately enabled, not a default.) ## Bootstrap / initial backfill When setting up a new source or rebuilding SAG from Git after an incident, this is the tool to use — still dry-run first, which is especially important since the number of files can be large (check `max_files` in the manifest, increase it explicitly if needed instead of letting it error out). ## Not for use with Publishing a single document you just wrote — use the `sag-publish` skill (via the `sag_publish` MCP tool), not batch sync. Sync is for: bootstrap, backfill, or periodic syncing of a large set at a human's explicit request.