Repository navigation
π Keep informational security notices separate from real human decision gates - #16032
Conversation
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
|
Integrated landed #16011 source95eafcdfa, preserving human_attention, canonical linked identities, Copy/Close hooks and head-aware waits; regenerated SPA from resolved source. Informational notices now also use informational accessible labels/tint and do not supply a blocking-security approval preface. A reviewer-mentioned login remains a quoted suggestion, never an assignment or exclusive authority; actor.login remains unknown unless independently supplied. Actual configured blocking labels retain their gate and warning. Exact-head full CI restarted on this pushed integration head; no decisions or live configuration changed. |
Open PR overlap checkBase branch: Conflicts are checked with every PR merged onto the current tip of
|
There was a problem hiding this comment.
Read the tree around the change; the core of this looks correct to me.
Why the derivation is sound. stageTrackerSentinel (src/pkg/dashboard/stage_tracker.go:1109-1151) only sets alert without the configured label when a bundle finding has f.Notice set β the trusted-author path its own doc-comment calls "comment, never the label". So SentinelNotice = SentinelAlert && !hasLabel at stage_tracker_build.go:406-408 is precisely the notice-only case, and the old decisionForPR blocker text ("will not merge until a person removes the sentinel-alert label", stage_tracker_decision_pr.go:146) was telling operators to remove a label that was never applied. Fixing that is a real correctness improvement, not just presentation. Linked-PR mirroring (stage_tracker_links.go:325) and both JS adapters carry the new field; every added static-src line is present in static/index.html. go build ./pkg/dashboard/ passes on f7906fb.
One stale comment (low). stage_tracker_build.go:409-412 still says the trusted-author notice "stays truthful for the gate word and the Decide⦠preface", but this PR removes notices from stageTrackerSentinelPreface (tracker-actions.js:1462) and from stageTrackerSentinelInLane. Worth updating the comment so the next reader isn't misled.
Open question, not a defect. stage_tracker_decision_pr.go:32 now also suppresses the gate when Reason is literally sentinel / ContributorLaneReasonSentinelBlocked as long as SentinelNotice is true. That requires a notice finding and a sentinel-blocked lane reason and no label β a contributor-lane block on a trusted-author PR β which I could not construct from the producers I read, so I'm noting it rather than claiming it. If it can't happen, fine; if it can, the lane reason should probably still win.
PR body matches the diff. Not approving β that is a human's signature.
β hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
Exact-head observer correctly stopped when Hive preparation advanced this branch from f7906fb to cbcddb4 at12:46:06Z. GitHub attributes this to app/hivecommons-hive merging current v5, not a local push. Compare shows current-base changes only; informational-notice source remains intact. Prior f790 verdict is not carried forward. Adopting/verifying the new head and owning its full fresh CI; no manual merge or deployment. |
There was a problem hiding this comment.
Read the Go and JS at cbcddb4 against the v5 tree rather than the diff alone. The change holds together:
stage_tracker.go:1137-1141is the only path that setsSentinelAlertwithout the live label (aNoticefinding in the verified bundle), so the derivation atstage_tracker_build.go:406-408(SentinelAlert && !label present) identifies exactly the trusted-author notice and nothing else.DecisionNeededis computed afterwards at:451, sodecisionForPRsees the flag.- In
decisionForPRthe reviewer-question case (stage_tracker_decision_pr.go:62) already outrankscase sentinel:(:105); the real effect of:32is on the latercase sentinel:at:145, which otherwise prints "will not merge until a person removes thesentinel-alertlabel" for a PR that does not carry that label. That blocker text was false for notice-only PRs; it is correct now, andTestStageTrackerSentinelNoticePreservesActualReviewerDecisionalso keeps it for the labelled case (:69-71). - The linked-PR copy at
stage_tracker_links.go:325and the two JS adapters intracker-row-layout.jsare the only propagation sites; the otherSentinelAlertassignment (api_repo_item_decide_delegated.go:687) is on the delegated-decision input struct, label-driven, and unaffected.
Looks correct to me.
One open question, not a defect: stageTrackerDecisionWhy (stage_tracker_actor.go:308) prefers the reviewer's question over humanizeStageReason(it.Reason) only when SentinelNotice is set. A non-notice PR at 085 with a reviewer question still gets the humanized lane reason first. If that narrowing is deliberate (keep this PR scoped to the notice case), fine; if the general case was also meant to quote the reviewer, the gate is one condition wider.
β hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
Hive status β updated in place; edits do not notify. Full history is in the Activity Monitor. Review findings from Read the tree at The defect this fixes is real. Checked and holding:
One suggestion, not a defect: @clubanderson this PR is held at Why:
Next: the hive does not push to your branch. Push a new head (or reply here); CI and the review swarm re-run automatically and the PR returns to the normal path. hive/awaiting-ci: rebased onto v5 (9616018 β f3db3d9) to clear the conflict in src/pkg/dashboard/static-src/js/tracker-row-layout.js and src/pkg/dashboard/static/index.html (linked-PR adapter field lists); kept both v5's new fields (state/terminal_merge/merged_by/human_attention) and the PR's sentinel_notice. The former merge commit f7906fb ('preserve landed human clarityβ¦') carried real changes (notice CSS, preface filter, chip aria-label, test asserts); it is preserved as a regular commit with the same author/message. Net PR diff unchanged. Authorship and sign-off unchanged. CI pending β sweep will merge when green. π Hive Agent: β hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 Evidence bundle `#16032f3db3d9` sha256:9a4f2b8ac2494d1f1e5d77fb74e8adf716143d8609e2f7c20869d5f80a82e4b8 β download from the dashboard or `hivectl review evidence #16032`.Review findings from Read this against the tree at
One note, not a defect: |
|
Prior immutable cbcddb4 reached all17 latest applicable workflow groups terminal success/skipped and all79 CheckRun contexts pass/skip. Hive then refreshed base again to9616018d9a06fe5c3f94662acc9306f3f60d7e1e, adding landed merge-toast0a4aa3d and Matrix API d5a7c57 via a bot merge. This is not a local push; I am preserving the canonical banner repo helper/asset and observing fresh CI on961, not labeling the new head green from cbc. |
|
FINAL immutable fresh-base CI receipt9616018d9a06fe5c3f94662acc9306f3f60d7e1e: all17 latest applicable workflow groups terminal SUCCESS/SKIPPED and all79 CheckRun contexts pass/skip, including full Go/Node/lint/docs/security/images and ARM. This is the refreshed head containing canonical landed banner repo helper and Matrix API, not prior cbc/f790 evidence. Owned worktree fast-forwarded/regenerated unchanged. No manual source merge or deployment by me; source remains Hive-managed. |
β¦ gates Signed-off-by: Andy Anderson <andy@clubanderson.com>
β¦bility Signed-off-by: Andy Anderson <andy@clubanderson.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
9616018 to
f3db3d9
Compare
|
@clubanderson this PR is back in your court (
Once the new head lands, CI and the review swarm re-run automatically and the PR returns to the normal path. |
|
GitHub force-push timeline attributes9616018βf3db3d92f0aff274c22a067eccdaa5a6371dd79a to hivecommons-hive at13:39:57Z. Rebased onto current v5; owned notice patch remains15 files/123 additions, no invented dependency on an unmerged feature. Fresh CI required onf3db; prior961 terminal79-check/17-group receipt remains immutable historical evidence only. |
|
merged by Hive trusted-author auto-merge: author @clubanderson holds role owner and GitHub permission admin on repo; CI green at f3db3d9. |
|
Thank you for your contribution! Your PR has been merged. We'd love to hear how your experience was: share feedback |
There was a problem hiding this comment.
Read the diff against the v5 tree plus the callers. This looks correct to me.
What I verified:
SentinelNotice(stage_tracker_build.go:406-408) isalert && !live-label. The only waystageTrackerSentinel(stage_tracker.go:1109-1150) promotesalertwithout the label is aNoticefinding in the verified bundle; the acknowledged path returns the label-onlyalert, so an acknowledged notice can never flip tosentinel_notice: true. Consistent with the label-only PR 109 / evidence PR 110 assertions added toTestStageTrackerSentinelFlag.- The new
!it.SentinelNotice &&guard indecisionForPR(stage_tracker_decision_pr.go:32) also covers the"sentinel"andcontributor-lane:sentinel-blockedreasons. Both are derived from the live label (automerge_contributor_lane.go:537-539vialabelBlockReason), so when either reason is present the label is present andSentinelNoticeis false β the guard cannot suppress a genuine lane gate. - JS side: with
stageTrackerSentinelInLane/AsSuffixnow false for notices,stageTrackerSentinelChipHtmlrenders the (info-tinted) chip instead of relying on the chevron word, so the notice stays visible at 085/071/072 as the body says.stageTrackerSentinelPrefaceexcluding notices matches the Go-side gate change. stageTrackerDecisionWhy(stage_tracker_actor.go:308-312) only re-orders reviewer-question-before-lane-reason whenSentinelNoticeis set, so non-notice rows are unchanged.
Body matches the diff; no scope beyond the stated change.
β hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
Final immutable source receipt: Hive auto-merged tested head f3db3d9 at2026-10-31T13:55:43Z as40d3be36f36525312dd039b0cccd6e4c40b7765d (app/hivecommons-hive). ARM completed after merge; final observed82 exact-head CheckRuns all pass/skip, all20 latest workflow/event groups success/skipped. DCO trailers verified, squash ancestor of freshly fetched origin/v5, canonical generation on the landed squash unchanged. Independent reviews caught and fixed exact-string adapter fixture plus prior human-attention integration; notice remains informational, real label/human decisions remain gates, reviewer-named person stays quoted provenance not exclusive authority. Owned worktree/local branch cleaned; remote branch auto-deletion verified. No deployment or policy activation. |
What changed
The tracker previously used
sentinel_alertfor both a real blocking label and a verified trusted-author informational notice. Decision rendering treated both as an open security gate.sentinel_noticeto the existing root/linked payload and canonical/inline adapters, using the existing verified observation and live configured label.Validation and boundaries
v5, independent of π fix: explain human requirements and optional tracker actionsΒ #16011/π fix: keep the preparation train connector continuous between adjacent membersΒ #16027. No live decisions, label changes, source-policy changes, deployment or activation.