🏥 Repository Health Dashboard

[github-actions]

🏥 Daily Health Check — 2026-05-26

Status: 🔴 2 critical · 🟡 2 warnings · 🔵 1 info
Since yesterday: 🆕 1 new · ✅ 3 resolved · 📌 4 unchanged

Summary: A welcome improvement — 3 findings resolved today, including the evaluation failure rate critical and both dotnet-blazor--coordinate-components and dotnet-test failures. The Markdown Linter startup_failure has recurred on main as predicted. The chronic dotnet-msbuild evaluation failure and eval duration critical remain active.

---

🆕 New Findings (1)

These appeared since the last health check (2026-05-25).

🟡 [P1] Markdown Linter workflow startup_failure on main

Fingerprint: pipeline:markdown-linter:startup:startup_failure
Severity: 🟡 Warning | First seen: 2026-05-26

The Markdown Linter workflow (.github/workflows/markdown-linter.lock.yml) had a startup_failure on main with 0 jobs executed.

• Run #23 — 2026-05-25T14:58:56Z (schedule, main)
• Conclusion: startup_failure — the agentic lock workflow failed before any jobs started
• Triggered by commit: 5810d492 ("test-anti-patterns / test-smell-detection: narrow-specialist re-pitch (Variant B)")
• 0 jobs ran; the workflow never started

Context: This finding was previously resolved on 2026-05-25 (no run in 24h window), but the prior investigation noted the root cause — super-linter/super-linter action not in .github/aw/actions-lock.json — was not fixed. As predicted, the failure recurred on the next scheduled run.

Suggested action: Add super-linter/super-linter to .github/aw/actions-lock.json (or equivalent action lock file) to resolve the startup failure permanently. 🔄 Investigation dispatched.

---

🔍 Investigation Results

Deep investigations are dispatched for new critical/warning findings.
The grooming workflow links results ~3 hours after this run.

Finding	Severity	Investigation	First Seen	Result
Evaluation failed — evaluate (dotnet-msbuild) / Run skill-validator	🔴 Critical	✅ Done	2026-05-20	The evaluate (dotnet-msbuild) job was killed by SIGTERM (exit code 143) after ~17 minutes, likely due to PR #673 adding a AITools.BinlogMcp MCP server with a dotnet dnx startup command that downloads a pinned prerelease from a custom NuGet feed at evaluation time; all 11 other evaluate jobs succeeded in the same run.
Evaluation avg duration — critical threshold exceeded	🔴 Critical	✅ Done	2026-05-14	The evaluate (dotnet-test) job now takes ~164 minutes per scheduled run due to the large number of complex agent evaluation scenarios in the dotnet-test plugin; this is documented growth, but it caused 7 consecutive daily run cancellations (May 7–13) and the workflow still operates within 16 min of the 180-min job timeout ceiling.
Markdown Linter workflow startup_failure on main	🟡 Warning	🔄 Dispatched	2026-05-26	⏳ Investigation dispatched — results arriving shortly...

---

✅ Resolved Since Yesterday (3)

These were in yesterday's report but are no longer detected.

🔴 Evaluation failed — evaluate (dotnet-test) / Run skill-validator

Fingerprint: pipeline:evaluation:evaluate-(dotnet-test):run-skill-validator:failure
Resolved: The evaluate (dotnet-test) job succeeded in today's scheduled evaluation run (run #3786), run time 2026-05-26T00:54:31Z. The runner preemption issue affecting dotnet-test appears to have been transient or the underlying contention resolved. The chronic evaluate (dotnet-msbuild) failure continues.

🔴 Evaluation failed — evaluate (dotnet-blazor--coordinate-components) / Run skill-validator

Fingerprint: pipeline:evaluation:evaluate-(dotnet-blazor--coordinate-components):run-skill-validator:failure
Resolved: No evaluate (dotnet-blazor--coordinate-components) job failure detected in the last 24h. The dotnet-blazor PR triggering this evaluation appears to have been merged or evaluation conditions changed. The evaluate (dotnet-blazor) job succeeded in run#3786.

🔴 Evaluation failure rate — Critical threshold exceeded

Fingerprint: pipeline:evaluation:failure-rate:critical
Resolved: Evaluation failure rate across all branches in the last 24h has dropped to ~14% (2 failures out of 14 non-cancelled runs), well below the Critical (>30%) and Warning (>15%) thresholds. The improvement is driven by the resolution of the dotnet-blazor--coordinate-components and dotnet-test failures — only the chronic dotnet-msbuild failure remains.

---

📌 Existing Findings (4)

These have been present since before today. Sorted by severity then age.

🔴 [P1] Evaluation failed — evaluate (dotnet-msbuild) / Run skill-validator · first seen 2026-05-20 · 7 occurrences

Fingerprint: pipeline:evaluation:evaluate-(dotnet-msbuild):run-skill-validator:failure
Severity: 🔴 Critical

The evaluate (dotnet-msbuild) job continues to fail at the Run skill-validator step in every scheduled evaluation run on main. This is the 7th consecutive day.

Today's run (run #3786):

• Started: 2026-05-26T00:54:31Z, concluded: 2026-05-26T03:03:29Z (~129 min total)
• Job evaluate (dotnet-msbuild): failed at Run skill-validator step
• All other evaluate jobs (dotnet, dotnet-test, dotnet-blazor, dotnet-aspnet, dotnet-data, dotnet-diag, dotnet-upgrade, dotnet-maui, dotnet-ai, dotnet-template-engine) succeeded

Root cause (from investigation): AITools.BinlogMcp MCP server startup causes resource exhaustion, leading to SIGTERM. The skill-validator evaluation job is killed while parallel vally-evaluate jobs succeed.

Suggested action: Remove or conditionally disable AITools.BinlogMcp during skill-validator evaluation. The BinlogMcp binary downloads at evaluation time from a custom NuGet feed, consuming resources that trigger runner preemption.

🔴 [P3] Evaluation avg duration ~129 min — critical threshold exceeded (>55 min) · first seen 2026-05-14 · 12 occurrences

Fingerprint: resource:eval-duration:critical
Severity: 🔴 Critical

Scheduled evaluation runs consistently exceed the 🔴 Critical threshold (>55 min). Today's scheduled run was 129 min.

Today's run (run #3786):

• Total run duration: ~129 min (00:54Z → 03:03Z)
• Note: run ended with a failure (dotnet-msbuild) but other jobs including dotnet-test completed successfully

7-day trend (scheduled runs):

Date	Duration
2026-05-26	~129 min
2026-05-25	124 min
2026-05-24	122.5 min
2026-05-23	119.5 min
2026-05-22	~131.7 min
7d avg	~125 min

Root cause (from investigation): The evaluate (dotnet-test) / vally-evaluate (dotnet-test) job now runs ~164 min of complex agent evaluation scenarios, dominating overall run duration.

Suggested action: Split the dotnet-test plugin evaluation into smaller scenario batches, or increase parallelism to bring total run time below 55 min.

🟡 [I8] Orphan plugin: dotnet-experimental not listed in marketplace.json · first seen 2026-05-14 · 12 occurrences

Fingerprint: infra:orphan-plugin:dotnet-experimental
Severity: 🟡 Warning

plugins/dotnet-experimental/ has a valid plugin.json but is not registered in .github/plugin/marketplace.json. The plugin is not discoverable by consumers.

• Plugin directory
• marketplace.json

Suggested action: Either register in all three marketplace files (.github/plugin/, .claude-plugin/, .cursor-plugin/) per CONTRIBUTING.md, or remove the directory if it is no longer active.

🔵 [I4] evaluation.yml uses --verdict-warn-only mode · first seen 2026-05-16 · 10 occurrences

Fingerprint: infra:verdict-warn-only
Severity: 🔵 Info

The evaluation workflow passes --verdict-warn-only to skill-validator. Evaluation failures are reported as warnings rather than hard PR-blocking errors. This is intentional configuration for the current rollout phase.

---

📊 Trends (7-day)

Metric	Today	7d Avg	Δ	Trend
Eval duration — scheduled (min)	129	~125	+4	↗️ Increasing (watch)
Eval success rate (main, scheduled)	0%	~0%	0%	➡️ Stable
Eval success rate (all branches, 24h)	~86%	~40%	+46%	✅ Improving
Eval scheduled cancellation rate	0%	0%	0%	➡️ Stable
Eval failure rate (all branches, 24h)	~14%	~45%	-31%	✅ Improving
Workflow failure rate (7d, main)	~7%	~15%	-8%	✅ Improving
Compute hours/day	~3.5h	~4.5h	-1h	✅ Improving

---

_{🤖 Generated by DevOps Health Check agentic workflow · Run #26432322026 · 2026-05-26T04:31Z UTC}

Generated by DevOps Daily Health Check · ● 1.7M · ◷

---

🔍 Investigation Results

Deep investigations are dispatched for new critical/warning findings.
The grooming workflow links results ~3 hours after this run.

Finding	Severity	Investigation	First Seen	Result
Evaluation failed — evaluate (dotnet-msbuild) / Run skill-validator	🔴 Critical	✅ Done	2026-05-20	The evaluate (dotnet-msbuild) job was killed by SIGTERM (exit code 143) after ~17 minutes, likely due to PR #673 adding a AITools.BinlogMcp MCP server with a dotnet dnx startup command that downloads a pinned prerelease from a custom NuGet feed at evaluation time; all 11 other evaluate jobs succeeded in the same run.
Evaluation avg duration — critical threshold exceeded	🔴 Critical	✅ Done	2026-05-14	The evaluate (dotnet-test) job now takes ~164 minutes per scheduled run due to the large number of complex agent evaluation scenarios in the dotnet-test plugin; this is documented growth, but it caused 7 consecutive daily run cancellations (May 7–13) and the workflow still operates within 16 min of the 180-min job timeout ceiling.
Markdown Linter workflow startup_failure on main	🟡 Warning	✅ Done	2026-05-26	The super-linter/super-linter action used in the super_linter job is missing from .github/aw/actions-lock.json, causing the gh-aw framework to reject the compiled workflow before any jobs start; fix requires adding the action entry to the lock file and recompiling.

Generated by DevOps Health — Groom Dashboard · ● 1.2M · ◷