Description
The Daily Regulatory Report (discussion #55519, 2026-08-24) found the Daily Performance Summary workflow (.github/workflows/daily-performance-summary.md) reported a 90-day rolling-window merged_prs count that swung from 76 to 42 (-45%), "issues resolved" from 49 to 15 (-69%), and open-issue backlog from 50 to 84 (+68%) between two runs one day apart — far larger than the ~1% drift a genuine 90-day rolling window should show day-to-day. The workflow queries via github-pr-query/github-issue-query with since: <90 days before this run> (.github/workflows/daily-performance-summary.md lines 92-120), which paginates via shared/github-queries-mcp-script.md's since-boundary logic — worth checking whether that pagination reliably captures the full window or silently truncates/varies run to run.
Expected Impact
A 90-day performance summary that swings 45-69% day-to-day is not trustworthy for tracking real backlog/velocity trends, defeating the purpose of the report. Fixing the window computation restores confidence in this daily metric.
Suggested Agent
New Agent — trace the since-based pagination path in shared/github-queries-mcp-script.md (github-pr-query/github-issue-query) as called from daily-performance-summary.md, and verify it returns a consistent result set for the same 90-day boundary across runs.
Estimated Effort
Medium (1-4 hours).
Data Source
DeepReport Intelligence Briefing analysis, 2026-08-25, sourced from Daily Regulatory Report discussion #55519 (Critical Issue 2).
Generated by 🔬 Deep Report · claude · agent · 249.5 AIC · ⌖ 8.79 AIC · ⊞ 12.4K · ◷
Description
The Daily Regulatory Report (discussion #55519, 2026-08-24) found the Daily Performance Summary workflow (
.github/workflows/daily-performance-summary.md) reported a 90-day rolling-windowmerged_prscount that swung from 76 to 42 (-45%), "issues resolved" from 49 to 15 (-69%), and open-issue backlog from 50 to 84 (+68%) between two runs one day apart — far larger than the ~1% drift a genuine 90-day rolling window should show day-to-day. The workflow queries viagithub-pr-query/github-issue-querywithsince: <90 days before this run>(.github/workflows/daily-performance-summary.mdlines 92-120), which paginates viashared/github-queries-mcp-script.md's since-boundary logic — worth checking whether that pagination reliably captures the full window or silently truncates/varies run to run.Expected Impact
A 90-day performance summary that swings 45-69% day-to-day is not trustworthy for tracking real backlog/velocity trends, defeating the purpose of the report. Fixing the window computation restores confidence in this daily metric.
Suggested Agent
New Agent — trace the
since-based pagination path inshared/github-queries-mcp-script.md(github-pr-query/github-issue-query) as called fromdaily-performance-summary.md, and verify it returns a consistent result set for the same 90-day boundary across runs.Estimated Effort
Medium (1-4 hours).
Data Source
DeepReport Intelligence Briefing analysis, 2026-08-25, sourced from Daily Regulatory Report discussion #55519 (Critical Issue 2).