[daily regulatory] Regulatory Report - 2026-08-23

[github-actions[bot]]

Reviewed 6 daily/rolling report discussions published in the last ~48 hours (Daily Performance Summary, Daily Code Metrics, Copilot Agent Analysis, Daily Team Evolution, Lockfile Statistics, DeepReport Intelligence). Overall data quality is good — most reports are internally consistent and reference clear time windows — but two discrepancies stand out: (1) total_workflows collapsed from 402 (2026-08-21 narrative) to 226 (2026-08-23 table) while Lockfile Statistics independently counts 290 .lock.yml files, a three-way mismatch on what should be the same underlying artifact set; and (2) Copilot Agent Analysis reports 52 agent-authored merged PRs in the same 24h window that Daily Team Evolution reports only 46 total merged PRs — agent-only merges exceeding the reported total is not possible and indicates a real counting bug in one of the two reports, not a scope difference.

No prior [daily regulatory] discussion was found in the discussions searched (most recent ~11-14 discussions), so there was nothing to supersede/close this cycle.

📋 Full Regulatory Report

📊 Reports Reviewed

Report	Title	Created	Status
Daily Performance Summary	[daily performance] Daily Performance Summary - 2026-08-23 (#55168)	2026-08-23T20:56Z	✅ Valid
Lockfile Statistics	[lockfile-stats] Lockfile Statistics — 2026-08-23 (#55163)	2026-08-23T20:22Z	✅ Valid
Daily Team Evolution	[daily-team-evolution] 🌱 Daily Team Evolution Insights - Aug 23 (#55160)	2026-08-23T20:19Z	⚠️ Issues
Daily Code Metrics	[daily-code-metrics] Daily Code Metrics Report - 2026-08-23 (#55139)	2026-08-23T18:48Z	⚠️ Issues
DeepReport (18:28Z run)	DeepReport Intelligence Briefing - 2026-08-23 (#55134)	2026-08-23T18:35Z	✅ Valid
Copilot Agent Analysis	[copilot-agent-analysis] Daily Copilot Agent Analysis - 2026-08-23 (#55127)	2026-08-23T18:25Z	⚠️ Issues
DeepReport (12:32Z run)	DeepReport Intelligence Briefing - 2026-08-23 (#55074)	2026-08-23T12:32Z	✅ Valid
Daily Performance Summary (prior day)	[daily performance] Daily Performance Summary - 2026-08-21 (#54617)	2026-08-21T20:55Z	✅ Valid
Lockfile Statistics (prior day)	[lockfile-stats] Lockfile Statistics — 2026-08-22 (#54908)	2026-08-22T20:22Z	✅ Valid
Daily Team Evolution (prior day)	[daily-team-evolution] 🌱 Daily Team Evolution Insights - 2026-08-22 (#54907)	2026-08-22T20:20Z	✅ Valid
Daily Code Metrics (prior day)	[daily-code-metrics] Daily Code Metrics Report - 2026-08-21 (#54595)	2026-08-21T18:57Z	✅ Valid

🔍 Data Consistency Analysis

Reference scratchpad/metrics-glossary.md for metric definitions and scopes. Note: the workflow-supplied glossary path was not reachable from the report-generation sandbox in this run, so this analysis used the repository's checked-in copy of the glossary for definitions — scopes below match that document.

Cross-Report Metrics Comparison

Metric	Daily Performance (90d window)	Copilot Agent Analysis (24h window)	Daily Team Evolution (24h window)	Scope Match	Status
Merged PRs (merged_prs)	76	52 (agent-authored only)	46 (all merges)	⚠️ Different scopes (90d vs 24h; agent-only vs all)	i️ See note below
Total PRs (total_prs)	100	61 (agent-authored, 24h)	not reported	⚠️ Different scopes	i️ Not comparable
Open Issues (open_issues)	50	n/a	n/a	—	✅ Single source

Metric	Daily Code Metrics 08-21	Daily Code Metrics 08-23	Lockfile Statistics 08-22	Lockfile Statistics 08-23	Scope Match	Status
Total workflows (total_workflows, .md sources)	402 (narrative text)	226 (table)	—	—	✅ Same scope (both claim .md workflow sources)	❌ Discrepancy
Lockfiles (.lock.yml)	—	226 ("Lock Files")	289	290	✅ Same scope (compiled .lock.yml count)	❌ Discrepancy vs Code Metrics

Scope Notes:

• merged_prs/total_prs: Daily Performance Summary uses a rolling 90-day window; Copilot Agent Analysis and Daily Team Evolution use a 24h rolling window — these are not directly comparable by design (see glossary "Cross-Report Comparison Guidelines").
• total_workflows vs lockfile count: per the glossary, total_workflows should equal the .lock.yml count 1:1 (every .md workflow source compiles to exactly one lock file), and the 08-23 Code Metrics report itself asserts this equivalence internally (226 .md = 226 .lock.yml). Lockfile Statistics independently counting 290 .lock.yml files on the same date is therefore a same-scope, same-day discrepancy of 64 files (~28%), not an intentional scope difference.

Consistency Score

• Overall Consistency: 82% (9 of 11 same-scope metric comparisons matched within tolerance)
• Critical Discrepancies: 1 (agent-only merged PRs exceeding total merged PRs in same window)
• Minor Discrepancies: 1 (workflow/lockfile count mismatch, total_workflows internal drop 402→226, and 226 vs 290 .lock.yml count same-day)

⚠️ Issues and Anomalies

Critical Issues

1. Agent-only merged PRs exceed total merged PRs in the same 24h window

   • Affected Reports: Copilot Agent Analysis ([copilot-agent-analysis] Daily Copilot Agent Analysis - 2026-08-23 #55127), Daily Team Evolution ([daily-team-evolution] 🌱 Daily Team Evolution Insights - August 23, 2026 #55160)
   • Metric: agent_prs_merged vs merged_prs
   • Description: Copilot Agent Analysis reports 52 Copilot-agent-authored PRs merged on 2026-08-23 (within a ~24h rolling window). Daily Team Evolution, covering the same 24h window (2026-08-22T20:14Z → 2026-08-23T20:14Z), reports only 46 total merged PRs across all contributors. Since agent-authored merges are a subset of total merges, 52 > 46 is logically impossible unless the two reports use materially different time windows or one has a counting error.
   • Expected: agent_prs_merged (52) ≤ merged_prs (46) for the same window.
   • Actual: agent_prs_merged (52) > merged_prs (46).
   • Scope Analysis: Both claim ~24h rolling windows ending around 2026-08-23T20:14–20:56Z; windows are close enough that this should not fully explain a 6-PR gap in the wrong direction.
   • Severity: High
   • Recommended Action: Audit Copilot Agent Analysis's PR-authorship detection (may be over-counting due to date-window drift or duplicate counting across its "last 3 days" table) and Daily Team Evolution's merge count (may be undercounting bot-authored commits). Reconcile against a single ground-truth query.

2. total_workflows metric collapsed from 402 to 226 while lockfile count independently shows 290

   • Affected Reports: Daily Code Metrics ([daily-code-metrics] Daily Code Metrics Report - 2026-08-21 #54595 on 08-21, [daily-code-metrics] Daily Code Metrics Report - 2026-08-23 #55139 on 08-23), Lockfile Statistics ([lockfile-stats] Lockfile Statistics — 2026-08-22 #54908 on 08-22, [lockfile-stats] Lockfile Statistics — 2026-08-23 (290 lockfiles, 41.1MB) #55163 on 08-23)
   • Metric: total_workflows (glossary: count of .md workflow source files in .github/workflows/)
   • Description: Daily Code Metrics' 08-21 narrative cites "402 markdown workflows," while its 08-23 table reports "Total Workflows (.md): 226" and "Lock Files (.lock.yml): 226" (asserting a 1:1 match). Lockfile Statistics, analyzing the same directory on the same dates, counts 289 (08-22) and 290 (08-23) .lock.yml files.
   • Expected: .md workflow count and .lock.yml count should match 1:1 per the glossary, and should not swing >40% day-over-day without an explanation (e.g., a mass file move/deletion).
   • Actual: 402 → 226 (08-21→08-23, Code Metrics) vs steady 289 → 290 (08-22→08-23, Lockfile Stats); the two reports disagree by 64 files (~28%) on the same day.
   • Scope Analysis: No scope difference is documented in the glossary for this metric — both reports claim to count the same underlying artifact set, so this is a true discrepancy, likely a measurement bug (possibly Code Metrics' file-count logic is filtering/miscounting, e.g., excluding subdirectories or hitting a glob issue).
   • Severity: Medium
   • Recommended Action: Reconcile Daily Code Metrics' workflow-counting logic against Lockfile Statistics' (which cross-checked against 290 physically parsed files) and correct whichever is wrong; add a sanity assertion (.md count == .lock.yml count) to Code Metrics to catch this automatically going forward.

Warnings

1. Internal percentage mismatch in Copilot Agent Analysis

   • Details: The summary line states "Merged: 52 (85.2%)" of 61 total PRs (52/61 = 85.2%, correct), but the report's own trend table lists "86.7%" as the 2026-08-23 success rate for the same 61/52 figures.
   • Impact: Minor internal inconsistency (85.2% vs 86.7%) suggests the trend table may use a slightly different denominator or a rounding/calculation slip; low impact but worth a quick fix for report credibility.

2. gh API authentication failures noted in two independent reports

   • Details: Copilot Agent Analysis ([copilot-agent-analysis] Daily Copilot Agent Analysis - 2026-08-23 #55127) reports gh API 401 errors preventing comment-count collection; DeepReport (DeepReport Intelligence Briefing - 2026-08-23 #55074, prior cycle) also references a CLAUDE_CODE_OAUTH_TOKEN silent-rejection chronic issue and other 401-related workflow failures mentioned via linked PRs (Fix Codex credential check 401 by disabling WebSocket transport #55106, Fix credentials-check workflow: replace unsupported --max-tokens flag for claude CLI #55095).
   • Impact: Recurring credential/auth issues across multiple daily workflows may be silently degrading data completeness (e.g., missing comment counts) without failing the workflow outright — worth a cross-workflow credential health check.

Data Quality Notes

• Discussion tooling in this environment enforces an ~11–14 discussion page-size ceiling per github-discussion-query call (larger requests error with "Argument list too long" in the underlying jq invocation), which limited this cycle's review to the 11 most recent discussions (spanning 2026-08-21 through 2026-08-23). No [daily regulatory] predecessor discussion was found in this window to close.
• Daily Team Evolution and Copilot Agent Analysis both self-report their own time windows precisely (ISO timestamps), which is good practice and made the cross-check above possible — other reports (e.g., Daily Performance Summary) should adopt the same precise-window disclosure.
• Lockfile Statistics is the most rigorously self-documented report reviewed, explicitly noting detection-methodology caveats (e.g., text-presence vs tool-wiring counts differing ~3x for create-discussion) — a good model for other reports to follow.

📈 Trend Analysis

Week-over-Week / Day-over-Day Comparison

Metric	2026-08-23	2026-08-21/22	Change
PRs merged (90d window, Daily Performance)	76	221 (change is expected: 90d window recalculated daily is not a fair day-over-day comparison)	n/a (window overlap, not additive)
Merged PRs (24h, Team Evolution)	46	30	+53%
Lockfiles parsed	290	289	+1 (+0.3%)
Agent PR success rate (Copilot Agent Analysis 3-day table)	86.7%	83.1% (08-22), 78.7% (08-21)	+8.0pp over 3 days
Avg agent PR duration	100 min	227 min (08-22), 169 min (08-21)	−56% vs prior day

Notable Trends

• Copilot coding agent throughput and reliability both improved together over the last 3 days (success rate up, average duration down) — a genuinely positive, internally-verified trend from Copilot Agent Analysis.
• Lockfile/workflow fleet size is essentially flat day-over-day per Lockfile Statistics (+1 file, +1.0% bytes), which conflicts with Code Metrics' much larger swing in total_workflows — reinforcing that the Code Metrics measurement (not the underlying repository state) is the likely source of the discrepancy flagged above.
• Both DeepReport cycles this period found no security-critical or systemic issues (0 DIFC integrity events, low firewall block rates), consistent with the generally healthy posture reflected in Daily Performance Summary and Lockfile Statistics.

📝 Per-Report Analysis

Daily Performance Summary (#55168, 2026-08-23)

Time Period: Rolling 90 days
Quality: ✅ Valid

Metric	Value	Validation
Total PRs	100	✅
Merged PRs	76	✅ (matches highlight bullet)
Open issues	50	✅ (consistent with "roughly half of 99" narrative)
Avg merge time	1.6h	✅

Notes: Internally consistent; narrative and tables agree.

Copilot Agent Analysis (#55127, 2026-08-23)

Time Period: Rolling 24h
Quality: ⚠️ Issues

Metric	Value	Validation
Agent PRs total	61	✅
Agent PRs merged	52	⚠️ Exceeds Team Evolution's total merged count (46) for same window
Success rate	85.2% (text) / 86.7% (table)	⚠️ Internal mismatch

Notes: Flag for follow-up per Critical Issue #1 above.

Daily Team Evolution (#55160, 2026-08-23)

Time Period: 2026-08-22T20:14Z → 2026-08-23T20:14Z
Quality: ⚠️ Issues

Metric	Value	Validation
Commits	30	✅
Merged PRs	46	⚠️ Lower than agent-only count from Copilot Agent Analysis (see Critical Issue #1)

Notes: Precise window disclosure is good practice; merge count needs reconciliation.

Daily Code Metrics (#55139, 2026-08-23)

Time Period: Point-in-time snapshot
Quality: ⚠️ Issues

Metric	Value	Validation
Total LOC	1,249,199	i️ Down from 1,916,807 on 08-21 (large swing, likely scope/methodology change — LOC composition also changed, e.g. YAML dropped from 649,909 to 13,764)
Total workflows (.md)	226	❌ Conflicts with Lockfile Statistics' 290 .lock.yml count same day (see Critical Issue #2)
Quality score	51.6/100	i️ Down sharply from 73.3/100 on 08-21

Notes: Multiple large swings between 08-21 and 08-23 snapshots (LOC, workflow count, quality score) suggest a methodology/scope change rather than genuine repository change of that magnitude in 2 days — recommend the report author document what changed in the counting methodology.

Lockfile Statistics (#55163, 2026-08-23)

Time Period: Point-in-time snapshot, compared to 08-22 prior snapshot
Quality: ✅ Valid

Metric	Value	Validation
Lockfiles parsed	290	✅ Consistent day-over-day trend (+1 from 289)
create_discussion tool-wired	91/290	✅ Self-consistent, methodology well documented

Notes: Best-documented report of the cycle; explicitly distinguishes text-presence counts from actual tool-wiring counts.

DeepReport Intelligence Briefing (#55134 and #55074, 2026-08-23)

Time Period: ~6h rolling cycles
Quality: ✅ Valid

Notes: Both cycles found no security-critical findings; identified two genuine data-quality bugs in other reporting workflows (engine-example counter miscounting, Delight CLI-quality section silently skipped) — consistent with the discrepancies flagged independently in this regulatory review, reinforcing that report-quality issues are a recognized, recurring theme across the daily-report fleet.

💡 Recommendations

Process Improvements

1. Add a .md-to-.lock.yml count assertion in Daily Code Metrics: Since these should always match 1:1 per the glossary, a simple equality check would have caught today's 226 vs 290 discrepancy automatically.
2. Standardize time-window disclosure: Require all daily reports to state their exact ISO-timestamp window (as Team Evolution and Copilot Agent Analysis already do) so cross-report comparisons like the one in Critical Issue rejig docs #1 are possible without guesswork.

Data Quality Actions

1. Reconcile the 52-vs-46 merged-PR discrepancy: Assign an owner to run a single ground-truth GraphQL/REST query for merged PRs in the exact window 2026-08-22T20:14Z–2026-08-23T20:14Z and compare against both reports' internal logic.
2. Investigate Daily Code Metrics' LOC/workflow-count methodology change between 08-21 and 08-23 (YAML LOC dropped ~98%, workflow count dropped ~44%) — likely a script change that should be documented, not a genuine 2-day repository shrinkage.

Workflow Suggestions

1. Cross-link a single "canonical metrics" cache-memory file (similar to Lockfile Statistics' 2026-08-22.json snapshot pattern) that other daily reports can read from for total_workflows/total_prs-style base facts, reducing duplicate (and inconsistent) computation across workflows.

📊 Regulatory Metrics

Metric	Value
Reports Reviewed	11
Reports Passed	7
Reports with Issues	4
Reports Failed	0
Overall Health Score	82%

---

Report generated automatically by the Daily Regulatory workflow
Data sources: Daily report discussions from github/gh-aw
Metric definitions: scratchpad/metrics-glossary.md

Generated by ⚖️ Daily Regulatory Report Generator · auto · 74.6 AIC · ⌖ 8.01 AIC · ⊞ 12.7K · ◷

• expires on Aug 26, 2026, 1:22 PM UTC-08:00

[github-actions[bot]]

This discussion has been marked as outdated by Daily Regulatory Report Generator.

A newer discussion is available at Discussion #55519.