You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[scanner] Contributor lifecycle notices as one upserted hive:contributor comment; ok-to-test ack as a reaction #18608
Part of #18450 (spam reduction phase 2). Items 5 and 6 of the proposal (~14/day): merge nudge, trust-wait, selfreport-seen, gfi, adopted and unpark edited into a single hive:contributor comment per issue; ok-to-test ack becomes a reaction, with a comment kept only for rejections.
Estimate and evidence: see the breakdown tables in #18450.
Looked at this one before writing code and stopped, because most of it is either already one comment per item or is a notice that has to notify.
What each notice does today
Notice
Today
Does a person need to act?
hive:contributor-nudge
One comment per PR, edited in place for new heads and the single repeat reminder (#17067)
Yes. It mentions the contributor ("back in your court"). An edit does not deliver a new mention, so folding it into a comment that already exists would drop that notification
hive:reporter-trust-wait
One per issue. Later scans do not repeat it
Yes. A maintainer has to add the label, and it tells the reporter why the issue waits
hive:gfi
One per issue, found by its marker and edited
Invitation to contributors. Already one comment
hive:unpark-notice:9879
Edited in place
"What to reply" ask. Already one comment
hive:adopted
One per adoption. It mentions the author on purpose ("push a new head here and say so"), and ParseAdoptionMarker needs the marker at the start of a comment
Yes. The author is losing ownership of the PR
hive:selfreport-seen
A new comment for every sighting on a fleet-report issue (hub and spoke-direct paths)
No. The issue body's affected-hives block already records hive, version and time
In the last 48 h sample, the only items carrying more than one of these markers were 4 PRs with adopted plus contributor-nudge (18482, 18510, 18514, 18536). Both of those mention the author deliberately. Merging them into one hive:contributor comment would save about 2 comments a day, and it would do that by dropping the notification on whichever notice came second. That breaks the rule for this epic that anything a person has to act on keeps a notifying comment. The ~8/day estimate in #18450 counted every lifecycle comment, not just the redundant ones.
ok-to-test. There is no separate acknowledgement comment. The /ok-to-test comment (<!-- hive-ok-to-test:<sha> -->) is the protocol command itself. Prow's trigger plugin and approve.sh read it from issue_comment, so a reaction cannot replace it. The hive-ok-to-test-rejected note is the human ask and stays. What does cut volume on Prow repos: once Prow has rejected the App's /ok-to-test in a repo, remember that for the repo and post the rejection guidance directly. That saves the App's command and Prow's "Cannot trigger testing" reply on every later fork PR (2 comments per PR, one of them Prow's).
Safe pieces that could still ship: (1) upsert hive:selfreport-seen into one comment per fleet-report issue (about 1/day), and (2) the repo-level Prow-rejection memory above. Leaving this open for a maintainer to choose the scope. Part of #18450.
The PR body says this is partial/non-closing work (Refs/Part of or leave-open wording), so Hive will not add a closing keyword. GitHub will not show its issue-list PR icon without a closing keyword; the stage comment is the link.
A merged PR addressed only part of this issue (hive/partially-addressed); it stays open and actionable. Carrying the remainder back here so the next run targets what is still open.
Owner decision: option A. Keep contributor-nudge, adopted and reporter-trust-wait as separate notifying comments; fold nothing further. Remaining scope: a repo-level memory that Prow refuses the App's /ok-to-test, so the first command is skipped on new fork PRs in known-refusing repos (survives restarts, TTL, clears when the self-contained hive-ok-to-test workflow is installed). The nudge-then-adopt duplicate is handled in #18872.
Closing: the proposed scope is done. #18748 upserted the selfreport-seen comment, #18838 covered the Prow rejection handling, and #18876 adds the repo-level memory that Prow refuses the App's /ok-to-test (fork lane skips the first command and posts the needs-human note directly; re-derived after restart; clears when hive-ok-to-test.yml is installed). Contributor-nudge, adopted and reporter-trust-wait comments stay separate by decision; the nudge/adopt ladder fix is in #18872.
Activity
clubanderson commentedon Nov 17, 2026
Claimed: an operator-dispatched agent is working on this now. Hive agents, please do not pick it up.
clubanderson commentedon Nov 17, 2026
Looked at this one before writing code and stopped, because most of it is either already one comment per item or is a notice that has to notify.
What each notice does today
hive:contributor-nudgehive:reporter-trust-waithive:gfihive:unpark-notice:9879hive:adoptedParseAdoptionMarkerneeds the marker at the start of a commenthive:selfreport-seenIn the last 48 h sample, the only items carrying more than one of these markers were 4 PRs with
adoptedpluscontributor-nudge(18482, 18510, 18514, 18536). Both of those mention the author deliberately. Merging them into onehive:contributorcomment would save about 2 comments a day, and it would do that by dropping the notification on whichever notice came second. That breaks the rule for this epic that anything a person has to act on keeps a notifying comment. The ~8/day estimate in #18450 counted every lifecycle comment, not just the redundant ones.ok-to-test. There is no separate acknowledgement comment. The
/ok-to-testcomment (<!-- hive-ok-to-test:<sha> -->) is the protocol command itself. Prow's trigger plugin andapprove.shread it fromissue_comment, so a reaction cannot replace it. Thehive-ok-to-test-rejectednote is the human ask and stays. What does cut volume on Prow repos: once Prow has rejected the App's/ok-to-testin a repo, remember that for the repo and post the rejection guidance directly. That saves the App's command and Prow's "Cannot trigger testing" reply on every later fork PR (2 comments per PR, one of them Prow's).Safe pieces that could still ship: (1) upsert
hive:selfreport-seeninto one comment per fleet-report issue (about 1/day), and (2) the repo-level Prow-rejection memory above. Leaving this open for a maintainer to choose the scope. Part of #18450.10 remaining items
hivecommons-hive commentedon Nov 17, 2026
Hive status — updated in place; edits do not notify. Full history is in the Activity Monitor.
PR #18748 covers this issue: #18748
The PR body says this is partial/non-closing work (
Refs/Part ofor leave-open wording), so Hive will not add a closing keyword. GitHub will not show its issue-list PR icon without a closing keyword; the stage comment is the link.🐛 Selfreport seen sightings as one upserted comment (#18748)
hivecommons-hive commentedon Nov 17, 2026
A merged PR addressed only part of this issue (
hive/partially-addressed); it stays open and actionable. Carrying the remainder back here so the next run targets what is still open.Remainder from the PR body:
Remaining: the repo-level Prow ok-to-test rejection memory proposed on the issue, and a maintainer decision on whether to fold any notifying notices.
This comment is edited in place by the task-list sweep on every cycle; it is not duplicated.
clubanderson commentedon Nov 18, 2026
Owner decision: option A. Keep contributor-nudge, adopted and reporter-trust-wait as separate notifying comments; fold nothing further. Remaining scope: a repo-level memory that Prow refuses the App's /ok-to-test, so the first command is skipped on new fork PRs in known-refusing repos (survives restarts, TTL, clears when the self-contained hive-ok-to-test workflow is installed). The nudge-then-adopt duplicate is handled in #18872.
🐛 Remember per repo that Prow refuses the App's /ok-to-test (#18608) (#…
clubanderson commentedon Nov 18, 2026
Closing: the proposed scope is done. #18748 upserted the selfreport-seen comment, #18838 covered the Prow rejection handling, and #18876 adds the repo-level memory that Prow refuses the App's /ok-to-test (fork lane skips the first command and posts the needs-human note directly; re-derived after restart; clears when hive-ok-to-test.yml is installed). Contributor-nudge, adopted and reporter-trust-wait comments stay separate by decision; the nudge/adopt ladder fix is in #18872.