You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Window evaluated: last 24 full hours (UTC), 2026-08-19T23:28:33Z to 2026-08-20T23:28:33Z
Total runs analyzed: 226
Detection-enabled runs: 205 (90.7%)
Regular runs: 21 (9.3%)
Misconfigured workflows found: 0
Note
No misconfigured workflows detected in this window. All four detection-misconfiguration rules (disabled despite >3 runs/7d, name implies detection role without the flag, detection-step failures, flip-flopping within the window) came back clean. Note the reporting window is effectively the entire available log history — the other 407 downloaded run directories had no aw_info.json/run_summary.json (empty stubs, likely activation-only or not-yet-materialized runs), so no data exists prior to 2026-08-20 to extend a 7-day check further back.
Comparison Chart
Rendered as two single-axis panels (run count, success rate) rather than a dual-axis overlay, with average token usage annotated directly on each bar.
Metric
Regular Runs
Detection Runs
Total runs
21
205
Success rate
81.0%
86.3%
Failure count
4
28
Avg tokens/run
19,420
24,484
Misconfigured Workflows
No misconfigured workflows detected in this window.
View All Run Metrics
75 distinct workflows produced runs in this window. Top entries by run count:
Workflow
Detection
Runs
Success %
Avg Tokens
Engine(s)
PR Sous Chef
yes
36
91.7%
55,227
pi
Test Quality Sentinel
yes
16
100.0%
9,024
copilot
PR Code Quality Reviewer
yes
16
100.0%
40,435
pi
Matt Pocock Skills Reviewer
yes
15
100.0%
16,184
copilot
Design Decision Gate 🏗️
yes
15
100.0%
14,310
claude
Impeccable Skills Reviewer
yes
15
100.0%
11,750
copilot
Ponytail Reviewer
yes
14
42.9%
3,324
copilot
PR Description Updater
yes
12
100.0%
9,242
copilot
Issue Monster
yes
7
100.0%
26,911
pi
AI Moderator
yes
5
0.0%
0
codex
Daily Go Test Parallelizer
yes
4
25.0%
89,014
copilot
Contribution Check
yes
3
100.0%
23,940
copilot
Avenger
yes
3
100.0%
9,201
claude
[aw] Failure Investigator (6h)
yes
2
100.0%
48,758
claude
Code Scanning Fixer
yes
2
50.0%
16,438
copilot
Smoke Copilot
no
2
0.0%
0
copilot
The remaining 59 workflows each had a single run in the window (mix of daily scheduled agents and smoke tests), spanning detection-enabled and regular configurations with no anomalies.
View Historical Trend
No trend chart available yet — the cache-memory trending store (/tmp/gh-aw/cache-memory/trending/detection-metrics/history.jsonl) held no prior entries, so today's metrics were recorded as the first data point. A 30-day trend chart will render automatically once ≥7 daily entries have accumulated.
Recommendations
Two workflows stand out for low success rates and warrant a look independent of detection config: AI Moderator (0% over 5 runs, e.g. run §32396393382) and Daily Go Test Parallelizer (25% over 4 runs, e.g. run §32393155021) — both have gh-aw-detection: true already, so this is a reliability issue, not a detection-config gap.
Ponytail Reviewer (42.9% success over 14 runs, e.g. run §32386403038) is the highest-volume workflow with a sub-50% success rate; worth checking for a flaky step before its next scheduled run.
Detection-enabled runs show ~26% higher average token usage than regular runs (24,484 vs. 19,420) — expected given the extra detection instrumentation, but worth tracking in the trend chart as the detection-enabled population grows.
No action needed on detection configuration itself this cycle — re-run this analysis after cache-memory accumulates a few more days of history to get a real trend line.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Summary
Note
No misconfigured workflows detected in this window. All four detection-misconfiguration rules (disabled despite >3 runs/7d, name implies detection role without the flag, detection-step failures, flip-flopping within the window) came back clean. Note the reporting window is effectively the entire available log history — the other 407 downloaded run directories had no
aw_info.json/run_summary.json(empty stubs, likely activation-only or not-yet-materialized runs), so no data exists prior to 2026-08-20 to extend a 7-day check further back.Comparison Chart
Rendered as two single-axis panels (run count, success rate) rather than a dual-axis overlay, with average token usage annotated directly on each bar.
Misconfigured Workflows
No misconfigured workflows detected in this window.
View All Run Metrics
75 distinct workflows produced runs in this window. Top entries by run count:
The remaining 59 workflows each had a single run in the window (mix of daily scheduled agents and smoke tests), spanning detection-enabled and regular configurations with no anomalies.
View Historical Trend
No trend chart available yet — the cache-memory trending store (
/tmp/gh-aw/cache-memory/trending/detection-metrics/history.jsonl) held no prior entries, so today's metrics were recorded as the first data point. A 30-day trend chart will render automatically once ≥7 daily entries have accumulated.Recommendations
gh-aw-detection: truealready, so this is a reliability issue, not a detection-config gap.References:
Warning
Firewall blocked 1 domain
The following domain was blocked by the firewall during workflow execution:
registry.npmjs.orgTo allow these domains, add them to the
network.allowedlist in your workflow frontmatter:See Network Configuration for more information.
All reactions