[habdelra]

Important

Merge order: #867 has merged; this PR targets main.

• The flags come from feat: announce a bulk realm write once, and let a deploy push without realm events boxel#6903 (realm push --no-realm-events) and feat: index public-realm deploys and system reindexes without realm events boxel#6914 (file write --no-realm-events). Each step checks the CLI's --help first, so this is safe to merge in any order relative to them. It takes effect once a latest boxel-cli with the flags is published (boxel-cli publishes on every boxel main merge) and the environment's realm server honors X-Boxel-Realm-Events: none; an older realm server ignores the header and the push still announces. No Merges after: line is needed, because nothing here breaks on an older boxel.

What this changes

A catalog deploy is system-initiated: no one with the catalog open asked for it. So the realm should index the new content without announcing it, and open tabs pick it up on their next load instead of resetting their loader mid-session.

• Deploy to staging (sync-to-workspace.yml): the .boxelignore write and the realm push pass --no-realm-events when the installed boxel-cli lists it.
• Push to submissions (push-to-submissions.yml): the same for the per-PR push to /submissions/.

A CLI without the flag announces every write.

Test plan

• The first staging deploy after feat: announce a bulk realm write once, and let a deploy push without realm events boxel#6903 and feat: index public-realm deploys and system reindexes without realm events boxel#6914 are on boxel main (staging runs boxel main): the push log shows the flag in effect, the catalog indexes the new files, and the catalog's realm event room gets no events from the push.

🤖 Generated with Claude Code

[github-actions]

Staging Submissions Preview

This PR's content is pushed to the staging submissions realm: https://realms-staging.stack.cards/submissions/

Changed folders:

• .github/

Updated at 2026-10-22 15:04:58 UTC for commit d1322e6. Shared realm: only this PR's changed files are pushed; files touched by multiple PRs reflect whichever pushed last, and deleted files are not removed.

[habdelra]

[Claude Code 🤖] This review covered four things: how the flag detection behaves under Actions' bash -e -o pipefail (empty arrays, and the herestring over the captured --help output), whether the help grep can match falsely, whether every boxel call in these workflows that writes to a realm passes the flag, and whether the description could trip Merges after: parsing. It did not review the CLI or realm-server halves of the flag.

No blocking issues. There is one stale justification comment, explained in the inline thread.

1. The comment on the staging push step still says every write notifies every open session and floods Matrix. See the thread on sync-to-workspace.yml.
2. Description nit: in "A CLI without the flag keeps announcing, as it does today.", "today" is temporal. "A CLI without the flag announces every write." says the same thing in evergreen form.

[habdelra]

[Claude Code 🤖] Once this flag is in effect, the comment on this step ("Each file the realm writes reindexes and notifies every open session, so re-sending the whole catalog floods the realm and Matrix.") overstates. The realm holds back only the Matrix broadcast, so a silent push still reindexes every file it sends but notifies no one. Please qualify it, for example:

# Each file the realm writes reindexes, and a push without
# --no-realm-events also notifies every open session, so re-sending the
# whole catalog floods the realm (and Matrix).

Regression (this change makes the comment overstate). Non-blocking.

[habdelra]

[Claude Code 🤖] Fixed in 7de754d. The comment now says each file reindexes, and only a push without --no-realm-events notifies open sessions. The description's 'as it does today' is reworded too.

[habdelra]

[Claude Code 🤖] Reopening. This PR doesn't depend on the shared realm event room that cardstack/boxel#6906 proposed, which was closed under ADR 011 (cardstack/boxel#6921). It only makes system-initiated indexing (public-realm deploys and syncs, the post-deploy boxel-ui reindex, Grafana reindexes, the OpenRouter sync) run without realm events, so open tabs aren't reset by indexing their users didn't ask for. That holds whichever way public-realm events are delivered, and it cuts the load on Synapse under any of the alternatives. Hassan wants it to land, and it was already approved. Reducing fan-out for public realms is being redesigned separately.