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
thirdshift keeps each reviewer's report beside the Session log #490
What the review said survives the session, on every Harness:
A place for the reports. For each session that runs the code review, the Sessions module names a place for the reviewers' reports that git ignores. The shared review instruction tells the agent to write them there.
The skill writes there. The thirdshift-code-review skill writes each axis's report, with its list of files read, to the place the Session prompt names, if it names one.
thirdshift keeps them. When the session ends, its Resume included, thirdshift keeps whatever reports are there beside the session's Session log. Each report shares the log's name, with the axis in place of the log's extension.
The same on every Harness, since the place comes from the prompt, not from the Harness.
No reports, no failure. A session that leaves no reports gets a progress line. thirdshift keeps the reports, but it doesn't read them or decide anything from them.
The glossary matches. The Session log entry in CONTEXT.md says a session that ran the review has its reviewers' reports kept beside it.
Acceptance criteria
Through the Sessions module's execution-test seam, with the session faked: a session that leaves both reports in the named place has them kept beside its Session log, each under the log's name with its axis.
A session that leaves no reports gets a progress line, and its Run carries on.
When a Resume ran the review, its reports are the ones kept.
The same holds whichever Harness the faked session stands for.
The prompts page shows the four prompts naming the place for the reports.
The skill writes each axis's report there, and only when the prompt names a place.
Activity
JacobStephens2 commentedon Oct 7, 2026
Closed by #512, merged into issue-486 by a thirdshift Merge run.