1
Batched Issue Processing
Should the pipeline process multiple issues per worker session? Current: single-issue sequential FIFO with 100% success rate at 15 issues.
2
Long-Running Implementation Handling
How should the pipeline handle implementations that exceed Claude CLI context window?
3
GitHub API Retry Strategy
What retry/backoff strategy for transient GitHub API failures (429, 502, timeouts)?
4
CSA Handoff Back-Link
Should completed ADW runs automatically update the CSA _handoff.json manifest?
5
Allowlist Scaling
How should the 22-repo allowlist scale? Manual curation becomes friction past 30 repos.