DeepReport Intelligence Briefing - 2026-07-21

[github-actions[bot]]

🔍 Executive Summary

The gh-aw fleet is healthy and stable: 93.1% workflow success (204 runs), token reporting fully recovered (11.03M captured), and safe-outputs back to 100% (55/55, 70 writes) after yesterday's 94.6% dip. The top finding is a real observability blind spot — the OTel agent.conclusion span reports run.status=success/STATUS_CODE_OK even when the Copilot CLI step exits non-zero (issue #47076, currently unlabeled), so failure-based alerting sees ~0 failures while the Actions UI shows dozens. Most urgent action: label and route #47076, and reverse the multi-day Copilot merge-rate slide (76.5% ▼, dragged by a WIP-prompt cluster merging at just 37.8%).

🚨 Top 5 Findings

• OTel silent-green (HIGH, otel: gh-aw.agent.conclusion reports run.status=success / STATUS_CODE_OK when the Copilot CLI step exits non-zero (error_count=0) #47076 — unlabeled). Conclusion span shows success/error_count=0 when the Copilot CLI step exits non-zero; residual gap after fix(otlp): derive gh-aw.run.status and status.code from output errors when conclusion env var is absent #33037/[otel-advisor] populate gh-aw.agent.conclusion span attribute on conclusion spans #35182. Real failures invisible in App Insights.
• Copilot merge rate sliding. 76.5% ▼ (-4.3% over 8 runs, from 80.8%); the wip/progress/started cluster (37 PRs) merges at only 37.8% — the single biggest throughput drag ([prompt-clustering] Copilot Prompt Clustering — 2026-07-21: merge rate 76.5% (▼), WIP prompts the top drag #47062).
• MCP telemetry near-blind. 18/20 sampled MCP runs emit neither gateway.jsonl nor rpc-messages.jsonl — 10% coverage; one failed run had no telemetry to diagnose ([observability] Observability Coverage Report - 2026-07-21 #46935).
• Health metrics overstate ill-health. All 14 audit failures were 0-turn/driver-exit (zero agent-logic/missing-tool/MCP); 3 reddened after the agent finished. Intentional credit-guardrail tests also shave ~1.5pts off prod-main by design ([audit-workflows] 🩺 Agentic Workflow Audit — 2026-07-20 (24h): HEALTHY 93.1% #46930).
• Positive: broad recovery. Token null-gap closed (11.03M), safe-outputs 100% with resilient failure-isolation ([safe-output-health] 🏥 Safe Output Health Report - 2026-07-21 #46981), pi 29/29 and codex 8/8 clean, weekly issues well-triaged (1/500 unlabeled, 0 open >7d). API steady at 90k today though the 7d avg is climbing (~118k).

✅ Actionable Agentic Tasks

Seven data-driven quick-win issues were filed this run (each verified non-duplicate against the open backlog):

1. Add console.Print* wrappers — Medium; kills the 1,640 two-step + 239 raw stderr calls ([repository-quality] Repository Quality Improvement Report — Console Print Wrapper Gap #47087).
2. Triage/label OTel bug otel: gh-aw.agent.conclusion reports run.status=success / STATUS_CODE_OK when the Copilot CLI step exits non-zero (error_count=0) #47076 — Fast; restores failure alerting (only unlabeled issue).
3. Exclude intentional credit-guardrail tests from prod-main rollups — Quick; recovers ~1.5pts of true health ([audit-workflows] 🩺 Agentic Workflow Audit — 2026-07-20 (24h): HEALTHY 93.1% #46930).
4. Investigate MCP telemetry coverage gap — Medium; closes the 10%-coverage blind spot ([observability] Observability Coverage Report - 2026-07-21 #46935).
5. Triage stale WIP-titled Copilot PRs — Medium; removes the 37.8% merge drag ([prompt-clustering] Copilot Prompt Clustering — 2026-07-21: merge rate 76.5% (▼), WIP prompts the top drag #47062).
6. Scope the Tidy workflow to cut spiky API bursts — Quick; 11k calls / 9 runs ([api-consumption] 📊 GitHub API Consumption Report — 2026-07-21 #47065).
7. Classify driver-exit vs agent-logic failures in health metrics — Medium; sibling to otel: gh-aw.agent.conclusion reports run.status=success / STATUS_CODE_OK when the Copilot CLI step exits non-zero (error_count=0) #47076 ([audit-workflows] 🩺 Agentic Workflow Audit — 2026-07-20 (24h): HEALTHY 93.1% #46930 rec rejig docs #1).

View supporting data

Metric	Value
Runs (24h)	204 (190 ✅ / 14 ❌) = 93.1%
Tokens captured	11,034,930 (null-gap closed)
Safe-output job success	100% (55/55), 70 writes
Copilot PR merge rate	76.5% ▼ (from 80.8%)
REST API calls (today)	90,320 · 7d rolling avg ~118k
Weekly issues (500 window)	103 open / 397 closed · 1 unlabeled · 0 open >7d
Discussions updated (7d)	50 (Audits 33, General 11, Announcements 5, Daily News 1)
Per-engine (audit)	pi 29/29 · codex 8/8 · copilot 120/129 (93%) · claude 31/35 (89%)

Sources analyzed: Agentic Workflow Audit (#46930), Safe Output Health (#46981), Prompt Clustering (#47062), API Consumption (#47065), Observability Coverage (#46935), Repository Quality (#47087), Sergo (#46979), Schema Consistency (#46991), LintMonster (#46956), Daily Status (#47051), prior DeepReport (#46873), weekly issues (500).

Continuity: prior DeepReport was 2026-07-20 (#46873); repo-memory before that was 3 months stale (2026-04-03). Memory refreshed this run (patterns, trends, flagged items). All schema/lint/doc-sync findings this cycle were already ticketed (#47013, #46990, #46866, #47014/#47083/#47084, #46514/#46955) and deliberately not re-filed.

References:

• §29843150600
• §29825753050
• §29789009312

Generated by 🔬 Deep Report · age00 200.9 AIC · ⌖ 17.2 AIC · ⊞ 9.9K · ◷

• expires on Jul 28, 2026, 7:27 AM UTC-08:00

[github-actions[bot]]

This discussion has been marked as outdated by Deep Report.

A newer discussion is available at Discussion #47350.