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), ending 2026-10-13T23:23Z
Run directories collected: 301 (1 GB budget, no limit reached)
Runs analyzed with usable detection-feature data (aw_info.json present): 218
Excluded from classification: 83 runs that were gated/skipped before the agentic engine started (no aw_info.json — e.g. skip-bots/skip-roles/rate-limit gate checks tripped in pre_activation), so detection status could not be determined for them
Detection-enabled runs: 215 (98.6%)
Regular runs (gh-aw-detection: false): 3 (1.4%)
Misconfigured workflows found: 0
Note
No misconfigured workflows were detected in this window. All three opted-out runs belong to one-off smoke-test workflows, and no detection-job failures or flag alternation were observed.
Comparison Chart
Metric
Regular Runs
Detection Runs
Run count
3
215
Success rate
100.0%
89.3%
Failure count
0
17
Avg. token usage
9,192
131,816
Detection-enabled runs use far more tokens on average and have a lower success rate — but this reflects workload mix (detection covers most of the fleet, including large reviewer/analysis workflows), not a cost of the detection feature itself. The 3 regular runs are all single-run smoke tests (Smoke Copilot Routed Claude, Smoke Copilot SDK Routed Claude, Smoke Pi Routed Claude), too few to compare fairly against 215 detection-enabled runs.
Misconfigured Workflows
No misconfigured workflows detected in this window.
Rule 1 (gh-aw-detection: false with >3 runs/7d): none of the 3 opted-out workflows exceeded 1 run even in this 24h sample; a full 7-day check is not possible since logs only cover the last 24h.
Rule 2 (detection job itself failing): no run had a detection job with a failing conclusion — all detection job runs observed were success or skipped (skipped consistent with the paired agent job also being skipped by gating logic, not a detection-step error).
Rule 3 (flag alternation within 24h): no workflow switched between gh-aw-detection: true/false/absent during the window.
(+28 more single-run workflows, each contributing 1 run)
Mostly Enabled
28
—
—
—
Remaining 28 workflows each had exactly 1 run in the window (mix of success/failure, all detection-enabled); omitted from this table for brevity but included in the aggregate metrics above.
View Historical Trend
Based on 5 of 16 recorded history points falling within the strict last-30-day window (2026-09-13 to 2026-10-13) — there is a logging gap between 2026-09-13 and 2026-10-07 in the trending cache.
Recommendations
No action required for detection configuration this window — no misconfigurations found.
Daily Go Test Parallelizer has a notably low 25% success rate (3 of 4 runs failed) across this window; worth investigating independently of detection status, as it's a recurring reliability signal rather than a detection issue.
Agentic Workflow Audit Agent and Daily Reliability Review each failed their single run in this window with very high token usage (1.14M and 647K tokens respectively) — worth a follow-up to confirm these aren't runaway-token failures before they recur.
Consider re-running this report after the trending cache's logging gap (2026-09-13 → 2026-10-07) is backfilled or stabilizes, so the 30-day trend chart reflects a fuller history.
References:
§38432161460 — Daily Go Test Parallelizer, lowest success rate (25%) in window
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
aw_info.jsonpresent): 218aw_info.json— e.g.skip-bots/skip-roles/rate-limit gate checks tripped inpre_activation), so detection status could not be determined for themgh-aw-detection: false): 3 (1.4%)Note
No misconfigured workflows were detected in this window. All three opted-out runs belong to one-off smoke-test workflows, and no detection-job failures or flag alternation were observed.
Comparison Chart
Detection-enabled runs use far more tokens on average and have a lower success rate — but this reflects workload mix (detection covers most of the fleet, including large reviewer/analysis workflows), not a cost of the detection feature itself. The 3 regular runs are all single-run smoke tests (
Smoke Copilot Routed Claude,Smoke Copilot SDK Routed Claude,Smoke Pi Routed Claude), too few to compare fairly against 215 detection-enabled runs.Misconfigured Workflows
No misconfigured workflows detected in this window.
gh-aw-detection: falsewith >3 runs/7d): none of the 3 opted-out workflows exceeded 1 run even in this 24h sample; a full 7-day check is not possible since logs only cover the last 24h.detectionjob with a failing conclusion — alldetectionjob runs observed weresuccessorskipped(skipped consistent with the pairedagentjob also being skipped by gating logic, not a detection-step error).gh-aw-detection: true/false/absent during the window.View All Run Metrics
Remaining 28 workflows each had exactly 1 run in the window (mix of success/failure, all detection-enabled); omitted from this table for brevity but included in the aggregate metrics above.
View Historical Trend
Based on 5 of 16 recorded history points falling within the strict last-30-day window (2026-09-13 to 2026-10-13) — there is a logging gap between 2026-09-13 and 2026-10-07 in the trending cache.
Recommendations
Daily Go Test Parallelizerhas a notably low 25% success rate (3 of 4 runs failed) across this window; worth investigating independently of detection status, as it's a recurring reliability signal rather than a detection issue.Agentic Workflow Audit AgentandDaily Reliability Revieweach failed their single run in this window with very high token usage (1.14M and 647K tokens respectively) — worth a follow-up to confirm these aren't runaway-token failures before they recur.References:
All reactions