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
bug(unic-archon-dlc): a review sub-agent that omits its Commands I ran section reads as one that ran no check #670
#660 has review in unic-dlc-pr-review.yaml ask both review sub-agents to end their report with a Commands I ran section, and record the checks listed there in reported_checks. The request binds nothing. A sub-agent that omits the section, or lists a command without an outcome, gives the same empty reported_checks as one that ran no check.
synthesize also drops any positive that says a check passes, and after #660review strips the Commands I ran section from the reports. So a sub-agent that ran a type-check and said so only in its prose, outside the section, leaves no trace in the summary: not as a fact, and not as an attributed claim. That is the shape of run 3's case in #487 § Observed, now silent rather than unattributed.
Acceptance criteria
review tells apart a sub-agent that listed none under Commands I ran from one whose report has no such section.
A missing section reaches the checks block as a sentence that names the axis and says its sub-agent did not report which commands it ran. It never reads as "no check ran".
A check result a sub-agent states in its prose outside the section is recorded in reported_checks as that sub-agent's claim, by the same rules as a listed check, and is never dropped without a trace.
The Box doc and ADR-0037 § Decision item 8 describe the change.
The pull request states that the change is unproven in the Consumer until a Consumer run reads it.
Filed by the /afk-loop 498 orchestrator from round 1 of the review of PR #669, the fix for #660.
What was reproduced, and how: by reading unic-dlc-pr-review.yaml at 8dd867fa, not by a run. The Commands I ran request in review sets no rule for a report that omits the section, and review has no instruction for that case, so the natural output is reported_checks: []. synthesize then drops a positive that claims a check passed. No Consumer run has exercised this.
Deferred from #660 because it is a Medium, not one of #487's criteria, and #660 lands without it. It depends on #660's reported_checks field, hence the blocked_by edge.
Closed as superseded by #678's resolution, approved by the maintainer on 2026-10-30: run 4 is dropped and unic-archon-dlc is frozen. Reason for this ticket: it carries one linked line to #687: an omitted section reads as "no check ran".
Activity
orioltf commentedon Oct 28, 2026
Filed by the
/afk-loop 498orchestrator from round 1 of the review of PR #669, the fix for #660.What was reproduced, and how: by reading
unic-dlc-pr-review.yamlat8dd867fa, not by a run. TheCommands I ranrequest inreviewsets no rule for a report that omits the section, andreviewhas no instruction for that case, so the natural output isreported_checks: [].synthesizethen drops a positive that claims a check passed. No Consumer run has exercised this.Deferred from #660 because it is a Medium, not one of #487's criteria, and #660 lands without it. It depends on #660's
reported_checksfield, hence theblocked_byedge.orioltf commentedon Oct 30, 2026
Closed as superseded by #678's resolution, approved by the maintainer on 2026-10-30: run 4 is dropped and
unic-archon-dlcis frozen. Reason for this ticket: it carries one linked line to #687: an omitted section reads as "no check ran".—
unic-sdlc-orchestrator-9