Repository navigation
π Activity Monitor: delegated adopt-supersede reports executing β adopted/could-not-adopt, routes workflow-path to the workflows blocker, re-adoption survives the mutation journal - #15206
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 |
There was a problem hiding this comment.
Intent alignment vs #15188 β (a) ledger executing/no-effect/blocked, (c) prior_copy re-adoption and (d) hover all match the issue; automerge_adoption_orphan.go:274 re-seeds AdoptedBy=dead before recoverOrphanedAdoption calls supersedeAdoptedPR, so the new prior_copy input (automerge_contributor_adopt.go:725-727) is genuinely distinct on the retry. One gap in (b):
Workflow-path rule is wider than the PR describes β src/pkg/delegatedecision/policy.go:463. The body says a PR "that would otherwise be contributor-offline is now capability-gap", and #15188 (b) asks to replace a doomed adopt-supersede. But the new rule has no !in.ContributorOnline guard, so an online contributor's .github/workflows/* PR with red CI / conflict / missing DCO β which previously fell through to fix-then-merge (policy.go:485-500, a direction the author can act on and that needs no App permission) β now parks as capability-gap recommend-only. That turns a working decision into "hive cannot clear this by deciding" for PRs the issue never mentions.
if !in.HiveAuthored && !in.ContributorOnline && in.WorkflowPath && (in.AdoptionStalled || in.Conflicting || !in.CIGreen || in.DCOMissing) {
Worth adding the inverse case to TestDecidePRRules (ContributorOnline: true, WorkflowPath: true, CIGreen: false β fix-then-merge) so the gate holds.
β hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
Open PR overlap checkBase branch: Conflicts are checked with every PR merged onto the current tip of
|
There was a problem hiding this comment.
Correctness review (read at head 2d8425d; Go unit tests could not be run here and did not run in CI β prow build/merge jobs are skipping because the PR is CONFLICTING against v5, so the new tests are unverified).
1. An executing adopt-supersede renders as failed "could not adopt: β¦" immediately, not "executingβ¦" β src/pkg/dashboard/operator_actions.go:638-639
operatorActionExecutionFor falls back to it.Reason whenever it has the adoption-stalled: prefix and treats it as a terminal refusal. But that prefix is the ordinary 085 park reason for a stalled adoption (automerge_contributor_adopt.go:1560-1568), and it is exactly what makes the policy answer adopt-supersede in the first place (pkg/delegatedecision/policy.go:472, in.AdoptionStalled). The tracker copies it onto the row (stage_tracker.go:1753, it.Reason = rec.Reason), so on the very next render after the click: Refused = "adoption-stalled:contributor-lane:ci-red" β operatorActionExecutingView second case (operator_actions.go:654-657) β State = failed, Progress = "could not adopt: β¦". The "executingβ¦" state the PR body promises is unreachable on the common path. Because the Refused cases precede AdoptedBy > 0 (:649-677) and MarkAdopted does not rewrite rec.Reason (stagelabels/store.go:1026), it also masks "adopted as #N" until the next Observe.
TestDelegatedAdoptSupersedeReportsExecutingUntilAdopted misses this because it builds items without Reason even though the store record carries adoption-stalled:contributor-lane:ci-red (api_repo_item_decide_delegated_test.go, new test, items := []StageTrackerItem{{Repo:β¦, Number: 8, Stage: 85, Head: "aaa"}}).
Only rec.AdoptEscalatedReason is a terminal refusal; drop the fallback:
if rec.AdoptEscalatedReason != "" && (rec.AdoptEscalatedHead == "" || it.Head == "" || rec.AdoptEscalatedHead == it.Head) {
out.Refused = rec.AdoptEscalatedReason
}
return out
}
and set Reason: "adoption-stalled:contributor-lane:ci-red" on the items in TestDelegatedAdoptSupersedeReportsExecutingUntilAdopted so the test covers the real row.
2. TestOperatorActionExecutionForReadsRecord fails as written β src/pkg/dashboard/operator_actions_test.go (new test, first assertion) vs operator_actions.go:629-631
The test calls operatorActionExecutionFor(it, stagelabels.Record{}, false) and requires got.Refused == it.Reason, but the function returns at if !hasRec { return out } before the it.Reason branch, so Refused == "" β t.Fatalf("no record: β¦"). With the fix in (1) the assertion should become got.Refused != "" || got.AdoptedBy != 0 (no record β no refusal).
- if got := operatorActionExecutionFor(it, stagelabels.Record{}, false); got.Refused != it.Reason || got.AdoptedBy != 0 {
+ if got := operatorActionExecutionFor(it, stagelabels.Record{}, false); got.Refused != "" || got.AdoptedBy != 0 {3. Rebase onto v5 so prow actually runs go test on the dashboard/automerge/delegatedecision packages before merge.
The automerge side checks out: recoverOrphanedAdoption keeps rec.AdoptedBy on the store when it calls supersedeAdoptedPR (automerge_adoption_orphan.go:200-205), so prior_copy is the dead copy as intended, and the marker path MarkAdopteds before recovering (:270-293).
β hive: agent=adjudicator-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
There was a problem hiding this comment.
The adoption-engine half of this looks correct to me; the PR is currently CONFLICTING against v5 and needs a rebase before the sweep can take it.
What I verified (src/pkg/github/automerge)
recoverOrphanedAdoption(automerge_adoption_orphan.go:200-205) callsstore.NoteAdoptionOrphanedand thensupersedeAdoptedPRwithout clearing the record;NoteAdoptionOrphaned(stagelabels/store.go:1158-1169) only setsAdoptOrphanedBy/Atand bumps the count, sostored.AdoptedByis still the dead copy whenautomerge_contributor_adopt.go:726reads it.prior_copyis therefore the dead copy's number on re-adoption and"0"on first adoption β thebranch_pushandpull_request_createclaims genuinely differ, which is what the journal needs to run the closure again.ClearAdoptiononly runs inrevertOrphanedAdoption(automerge_adoption_orphan.go:221), i.e. after the re-adoption attempt, so there is no path where the second attempt seesAdoptedBy == 0and collides with the first claim.replayedAdoptionCopy(:1691-1709) only reuses anopenPR and errors on a closed one, so a replay can never silently mark a dead copy as adopted;TestAdoptionReadoptsThroughReplayingJournalandTestAdoptionReplayedOpenCopyIsReusedcover both branches.
Not verified: the dashboard ledger/state changes (operator_actions.go, api_repo_item_decide_delegated.go, decision_glossary.go, policy.go, index.html) β I read the automerge path only. CI is green on this head apart from tide, which is the merge conflict.
β hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
2d8425d to
c974dc3
Compare
|
@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. |
There was a problem hiding this comment.
Conclusion: this PR's decidePR workflow-path rule overlaps with #15208 (scanner, child (b) of #15188) in a way that conflicts rather than duplicates, and after a rebase onto current v5 the rule as placed here misses the grace-elapsed path it is meant to close. The ledger/β-state and mutation-journal parts of this PR are not in #15208 and still stand on their own.
1. Name collision with #15208 β both cannot land as written.
This PR adds a field Input.WorkflowPath bool (src/pkg/delegatedecision/policy.go, Input struct, this diff). #15208 adds a method func (in Input) WorkflowPath() bool on the same struct, fed by a new Input.ChangedFiles []string plus the adoption-stalled:contributor-lane:workflow-path slug in in.Reasons. Go does not allow a field and a method with the same name, so whichever merges second fails to compile. Severity: high (build break), consequence: the second PR needs a rewrite of its policy hunk, not just a conflict resolution.
2. Ordering gap once rebased: grace-elapsed workflow-path heads still get adopt-supersede.
This diff's base predates the adopt-now rule (#15038) β the hunk context contains no AdoptionGraceElapsed. On current v5, decidePR evaluates adopt-now at src/pkg/delegatedecision/policy.go:463 before contributor-offline at policy.go:471. This PR inserts its capability-gap rule immediately above contributor-offline and conditions it on !in.ContributorOnline && (β¦), so a .github/workflows/ head whose grace has elapsed returns adopt-now β adopt-supersede β exactly the decided-but-never-executed outcome the body describes for #14822. #15208 places its rule above adopt-now and gates on in.AdoptionGraceElapsed || (!in.ContributorOnline && β¦), with a test "grace-elapsed workflow-path PR gets capability-gap, not adopt-now". Severity: medium; it would only show after rebase, which needs-rebase says is coming.
3. File-list source (minor, informational).
WorkflowPath here is derived from the existing single-page api.PullRequests.ListFiles(β¦ PerPage: delegatedDecisionGatherLimit) call (src/pkg/dashboard/api_repo_item_decide_delegated.go:610, limit 100, errors swallowed), so a >100-file PR with the workflow edit beyond page one is only caught via the park-reason slug. #15208 fills ChangedFiles from the paginated github.ListPRChangedFiles and fails closed on error (delegated_pr_context.go hunk).
Suggestion: let #15208 own the delegatedecision rule (it is narrower, already covers the grace-elapsed branch and the complete file list), rebase this PR on top of it, drop the WorkflowPath bool field/rule and RuleCapabilityGap rename here, and keep the operator-action ledger, delegatedStuckWhy, glossary workflow-path entry and re-adoption journal fixes, which #15208 does not touch. Alternatively fold #15208's three policy hunks into this PR and close #15208 β either way only one of the two WorkflowPath definitions should survive. I have not changed or closed anything.
β hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
There was a problem hiding this comment.
Style / maintainability review (head 314e9aa)
1. capability-gap is hard-wired to the workflows blocker, but the rule is generic β src/pkg/dashboard/operator_actions.go:556-557 and :789-793
delegatedVerdictOutcome maps any rule == RuleCapabilityGap to Plain(GlossaryWorkflowPath) ("the PR edits .github/workflowsβ¦"), and the blocked branch of stageTrackerOperatorActionExec unconditionally sets Blocker = capability:workflows-permission and Progress = "could not adopt: β¦". The same rule is also returned by decideIssue (src/pkg/delegatedecision/policy.go:359-360) whenever capabilityGapRE (policy.go:237, matches any "App β¦ lacks β¦ permission(s)" prose) fires on an issue. An issue row decided capability-gap will now hover "could not adopt: the PR edits .github/workflowsβ¦" β wrong noun, wrong verb, wrong blocker. Keep the ledger's blocked view driven by the record's own detail instead of a hard-coded workflow sentence:
case a.Outcome == operatorActionOutcomeBlocked:
out.State = operatorActionStateBlocked
- out.Blocker = operatorActionBlockerWorkflows
- out.Progress = "could not adopt: " + delegatedecision.Plain(delegatedecision.GlossaryWorkflowPath)
- out.Human = "Grant the hive's GitHub App the `workflows` permission and re-run the decision, or rebase and merge it yourself on GitHub, or close it as superseded."
+ out.Progress = a.Detail
+ out.Human = operatorActionHumanGrantPermission
+ if strings.Contains(a.Detail, "workflows") {
+ out.Blocker = operatorActionBlockerWorkflows
+ out.Progress = "could not adopt: " + delegatedecision.Plain(delegatedecision.GlossaryWorkflowPath)
+ out.Human = operatorActionHumanWorkflows
+ }and in delegatedVerdictOutcome use Plain(GlossaryWorkflowPath) only when the decision's reason names the workflow path (pass Decision.Reason/in.WorkflowPath through, or fall back to Plain(RuleCapabilityGap) for the issue-side rule).
2. Duplicated human-step sentence β operator_actions.go:682 and :793 carry the identical 150-char "Grant the hive's GitHub App the workflows permissionβ¦" string (a third variant is the glossary Detail at decision_glossary.go:174-176). Hoist it to a const operatorActionHumanWorkflows next to operatorActionBlockerWorkflows (:88-91) so the two views cannot drift.
3. Two prefixes for one meaning β delegatedDecidedDetail emits "hive decided: " (operator_actions.go:541,543) while delegatedVerdictOutcome emits "decided: " (:557,562) yet "hive decided: " for the human-merge path (:560). The SPA then needs a second, byte-identical regex branch (static/index.html:15332) to render both the same way. Pick one prefix on the Go side and drop the extra JS branch, or collapse the two JS lines into /^(?:hive )?decided: /.
4. RuleCapabilityGap lives in the "Labels the delegate writes or reads" const block β src/pkg/delegatedecision/policy.go:126-132. It is a rule slug, not a label; move it to its own block (or beside the other rule strings) so the block's doc comment stays true.
β hive: agent=ci-maintainer backend=copilot model=claude-fable-5.1 copilot=1.0.88
There was a problem hiding this comment.
Intent alignment vs #15188 (read at 314e9aa): the diff delivers (a), (b)/policy and (d) of the issue's fix list on the production (async) path; (c) is correctly attributed to the already-merged #15209. Two advisories, neither blocking:
-
Sync response path still settles a dedup
no-effectasexecuting/applied.settledDelegatedOutcome(src/pkg/dashboard/operator_actions.go:526) handlesrowActionNoEffect, but the mirror inoperatorActionOutcome(operator_actions.go:378-389) only checksrowActionNoopbefore falling intodelegatedVerdictOutcome(:393-397). Production runs withdelegatedDecisionRunAsync = true(api_repo_item_decide_delegated.go:67) so the ledger is written bySettle, but when the flag is off (tests, or any future sync caller) the second click on an adopt-supersede recordsexecutingagain β the exact "done while nothing adopted" shape the issue describes.--- a/src/pkg/dashboard/operator_actions.go +++ b/src/pkg/dashboard/operator_actions.go @@ func operatorActionOutcome(status int, body []byte) if res, ok := resp["result"].(map[string]any); ok { - if action, _ := res["action"].(string); action == rowActionNoop { + if action, _ := res["action"].(string); action == rowActionNoop || action == rowActionNoEffect { verdict := "" if d, ok := resp["decision"].(map[string]any); ok { verdict, _ = d["verdict"].(string) } var rr RowActionResult if raw, err := json.Marshal(res); err == nil { _ = json.Unmarshal(raw, &rr) } + if action == rowActionNoEffect { + return operatorActionOutcomeNoEffect, operatorActionNoEffectDetail(&rr), head, true + } return operatorActionOutcomeNoOp, operatorActionNoopDetail(verdict, &rr), head, true } }
-
Title and the "What changes β (b) Adoption worker" / Tests sections describe code that is not in this diff.
gh pr difftouches nopkg/github/automergefile and contains neitherTestAdoptionReadoptsThroughReplayingJournalnorTestAdoptionReplayedOpenCopyIsReused; the "Supersedes #15208 Β· layers on #15209" section already says they were dropped in favour of #15209. Suggest dropping "re-adoption survives the mutation journal" from the title and the (b) block + those two test names from the body so the squash message matches the change (the changelog fragment already gets this right).
β hive: agent=adjudicator-4 backend=copilot model=claude-fable-5.1 copilot=1.0.88
There was a problem hiding this comment.
Docs currency β two operator-facing docs enumerate the vocabulary this PR extends and were not updated (reviewed at 314e9aa).
src/docs/api-reference.md:286still documents the ledger asoutcome: pending|applied|no-op|failedandstate(waiting|addressed|stuck|pending|failed). This PR adds outcomesexecuting/no-effect/blocked(src/pkg/dashboard/operator_actions.go:60-73), statesexecuting/blocked(:85-86) and payload fieldsprogress/blocker/human(:628-634);dashboard/openapi.jsonwas updated for all of them, the prose row was not.
-outcome: pending|applied|no-op|failed, detail, head_sha, stage}`; `last_operator_action` adds the derived `state` (`waiting|addressed|stuck|pending|failed`), `stuck_after_seconds`
+outcome: pending|applied|no-op|failed|executing|no-effect|blocked|superseded-by-state-change, detail, head_sha, stage}`; `last_operator_action` adds the derived `state` (`waiting|addressed|stuck|pending|failed|executing|blocked`), `progress` / `blocker` / `human` (what the worker has done with an `executing` decision, the capability-blocker slug holding a `blocked` one β e.g. `capability:workflows-permission` β and what a person can do), `stuck_after_seconds`src/docs/labels-and-control-signals.md:1431("Decision glossary") promises every slug "has one plain-words entry here", but the Reasons table (:1484onward) has none of the three entriesdecision_glossary.go:174-180adds. No test syncs that table (only theindex.htmlmirror is checked,stage_tracker_glossary_test.go:17), so it needs a hand edit:
| `adoption-disabled` | adoption is switched off on this hive | Stage adoption is disabled on this hive (governor.labels.stage.adoption), so a request to take a PR over is never carried out. |
+| `adoption-executing` | the adoption engine is carrying it out | The decision is recorded and the adoption engine picks it up on its next stage pass (a few minutes): it pushes the commits to a hive branch and opens the hive's copy. Nothing to do yet. |
+| `adoption-stalled:contributor-lane:workflow-path`, `capability:workflows-permission`, `contributor-lane:workflow-path`, `workflow-path` | the PR edits .github/workflows; the hive's App lacks the `workflows` permission, so it cannot adopt it | The PR changes files under .github/workflows. The hive's GitHub App has no `workflows` permission, so the adoption fence refuses to publish those edits from the hive. Grant the App the `workflows` permission and re-run the decision, or have a person rebase and merge it on GitHub, or close it as superseded. |
+| `adoption-undone` | the hive's copy was closed without merging and the adoption was undone | The hive opened its copy, but the copy was closed without merging and a fresh one could not be opened, so the adoption was undone and the PR is back in its normal flow. Re-run the decision to adopt it again, or act on GitHub. |Out of scope (pre-existing): the same Rules table is also missing capability-gap (#15127) and human-merge (#15078).
β hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
β¦dopted or could-not-adopt, routes workflow-path PRs to the workflows capability blocker
Let hive decide β adopt-supersede settled the operator-action ledger as
applied ("done") as soon as the adoption/supersede request label landed,
and the β hover kept saying so while nothing happened.
- #14822 edits .github/workflows: the adoption engine's
fence refused the supersession (adoption-stalled:contributor-lane:
workflow-path) and the row fell back to a bare Needs human decision.
The policy now decides such PRs as capability-gap (recommend-only),
the ledger records blocked, and the badge names the
capability:workflows-permission blocker with what a person can do.
- #14947: the re-adoption after the copy #14977 was
closed was replayed by the mutation journal; that fix landed in
#15209 (prior_copy claim input, replayed provenance
resolved). This PR layers the ledger / hover states over it.
- settledDelegatedOutcome: adopt-supersede β executing, dedup β no-effect,
capability-gap β blocked. The tracker derives the β state from the
stage record (adopted as #N / could not adopt / still executing / stuck)
and exposes progress, blocker and human to the SPA hover.
Fixes #15188
Signed-off-by: Andrew Anderson <andy@clubanderson.com>
314e9aa to
1b65801
Compare
|
Thank you for your contribution! Your PR has been merged. We'd love to hear how your experience was: share feedback |
β¦he glossary tables Review follow-up for #15259: the human-merge Reasons row now carries its glossary aliases (author-association, contributor-lane:author-association, ready-for-human-merge, untrusted-author) like every other aliased row, and the three Reasons #15206 added to decision_glossary.go β workflow-path (with its aliases), adoption-executing and adoption-undone β get rows so the table matches the code on v5. Hive-Fix-Attempt: 1/6 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: ci-maintainer <280983584+hivecommons-hive[bot]@users.noreply.github.com>
β¦merge (#15259) * π docs: add capability-gap and human-merge rows to the decision glossary Rules table Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com> * π docs: list human-merge aliases and the #15206 adoption reasons in the glossary tables Review follow-up for #15259: the human-merge Reasons row now carries its glossary aliases (author-association, contributor-lane:author-association, ready-for-human-merge, untrusted-author) like every other aliased row, and the three Reasons #15206 added to decision_glossary.go β workflow-path (with its aliases), adoption-executing and adoption-undone β get rows so the table matches the code on v5. Hive-Fix-Attempt: 1/6 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: ci-maintainer <280983584+hivecommons-hive[bot]@users.noreply.github.com> --------- Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com> Signed-off-by: ci-maintainer <280983584+hivecommons-hive[bot]@users.noreply.github.com> Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Fixes #15188
What was wrong
"Let hive decide" β
adopt-supersedesettled the operator-action ledger asapplied("done β the hive decided: adopt-supersede (rule: contributor offline)") the moment theadoption/supersederequest label landed. Adding the label is not the adoption; the badge then said "done" while the row sat at 085.Two live failure shapes (log evidence in #15188):
.github/workflows/v2-ci.yml. The adoption engine's fence refused the supersession (adoption-stalled:contributor-lane:workflow-path,<!-- hive:adoption-workflow-refused -->) and the row fell back to a bare Needs human decision. The π fix: Let hive decide keeps capability-gap needs-human parkedΒ #15127capability-gaprule only regex-matched prose, not the fence slug or the file list.branch_push/pull_request_createclaims had identical inputs{base, supersedes, head_sha}to the first attempt, so the mutation journal replayed them as idempotent success without running the closure βcreated == nilβadoption failed to open App PR error=Noneβre-adoption did not open a fresh copy; reverting adoption.What changes
(a) Ledger / β state (
pkg/dashboard/operator_actions.go)settledDelegatedOutcome: adopt-supersede βexecuting("decided: adopt-supersede (rule) Β· executingβ¦"); dedup βno-effect;capability-gapβblocked; failed/no-op/applied as before.stageTrackerOperatorActionExecderives anexecutingrecord's state from the stage record:AdoptedBy>0β addressed "adopted as #N"; workflow-path refusal βblockedwithblocker: capability:workflows-permission(the slug π Delegated decision on a capability gap: model the blocker, stop the un-park/re-park loop, name the missing App permissionΒ #15134 uses, so it composes withpkg/blockerwhen that lands); other refusal / copy closed twice / copy failed βfailed"could not adopt: β¦"; copy closed β still executing "opening a fresh one"; nothing on record past the window β stuck.progress,blocker,human(openapi updated).(a) Policy (
pkg/delegatedecision)Input.WorkflowPath; a contributor PR under.github/workflows/that would otherwise becontributor-offlineis nowcapability-gaprecommend-only (keeps the park). The gather sets it from the PR's file list and fromcontributor-lane:workflow-pathin the reason.workflow-path(+ aliases incl.capability:workflows-permission),adoption-executing,adoption-undone; JS mirror regenerated.(b) Adoption worker (
pkg/github/automerge/automerge_contributor_adopt.go)prior_copyso a re-adoption after a closed copy is a distinct logical effect.pull_request_createwithout running the closure, the provenance URL is resolved; an open copy is reused, a closed one fails loudly instead of "None".(c) SPA hover (
static/index.html)executing(β³, dotted) andblocked(β , amber dashed); first line says what the hive is doing now / why it is blocked; "You can: β¦" fromhuman; options shown for blocked too.Tests
TestAdoptionReadoptsThroughReplayingJournal,TestAdoptionReplayedOpenCopyIsReused(replaying boundary; docs: add dashboard screenshot + qwen/deepseek cloud model boxes in SVGΒ #91 opened after docs: split SVG backends into Native/Aggregate, update READMEΒ #90 closed).TestDelegatedAdoptSupersedeReportsExecutingUntilAdopteddrives decision β ledgerexecutingβMarkAdoptedβ "adopted as docs: add dashboard screenshot + qwen/deepseek cloud model boxes in SVGΒ #91" β escalation βblocked.TestDelegatedWorkflowPathIsCapabilityGap, policy case,TestSettledDelegatedOutcomeDistinguishesExecution,TestStageTrackerOperatorActionExecutingFollowsTheWorker,TestOperatorActionExecutionForReadsRecord.Out of scope (noted in #15188): the reviewer
requires_humanroute not counting the delegated comment as an answer.Supersedes #15208 Β· layers on #15209
The scanner opened two child PRs off #15188 (#15203 β #15208, #15204 β #15209) three minutes after this one.
capability-gap) is covered here and goes further:capability-gapinstead of a doomedadopt-supersedeInput.ChangedFiles+Input.WorkflowPath()scanning the file list andReasons; gated on!ContributorOnline; fills files ingatherDelegatedPRContextInput.WorkflowPathbool set from the PR file list ingatherDelegatedInput(already fetched forLowRiskPaths) and from aβ¦:workflow-pathpark reason; same!ContributorOnlinegate (folded in from #15208);RuleCapabilityGapconstant shared with the issue-side ruleblocked, routed to thecapability:workflows-permissionblocker with the human stepexecutingβadopted as #N/could not adopt: β¦on the β hover, SPA states,openapi.jsonTestDelegatedAdoptSupersedeReportsExecutingUntilAdopted,TestDelegatedWorkflowPathIsCapabilityGap,TestOperatorActionExecutionForReadsRecordRebased on v5 after #15201 (durable supersede-now executor) and #15209 merged.