[audit-workflows] π Agentic Workflow Audit β 2026-07-17 (89.3% healthy, 19 fails all 0-turn copilot driver) #46310
Closed
Replies: 1 comment
-
|
This discussion has been marked as outdated by Agentic Workflow Audit Agent. A newer discussion is available at Discussion #46492. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
π Agentic Workflow Audit β 2026-07-17
Window: 07:46β21:07Z (~13.4h partial) Β· Runs: 177 Β· Success: 158 (89.3%) Β· Failures: 19
First audit in 11 days (prior: 2026-07-06). Health remains on the established 86β89% baseline β no novel anomalies; every failure maps to a known issue.
Key metrics
π Trend charts
Workflow Health (30d)

Success rate holds at 89.3%, squarely on the multi-week 86β89% band. Failures stay concentrated in copilot's 0-turn driver family; pi/claude/codex remain near-perfect. No regression versus the 07-06 window.
Fleet Token Usage (days with reported data)

Today's 10.67M reflects a fuller
metrics.TokenUsagecapture, not a real spike β most prior days report null due to the standing token-reporting gap, so day-to-day token comparison stays unreliable. AI-credits remains the trustworthy cost signal.1. Smoke CI β 6Γ 0-turn fail on push (chronic, recurrence β 7)
All 6 copilot/claude-sonnet-4.6
Execute GitHub Copilot CLIfailures at 0-turn/0-tok onpushto main. Known issuesmoke-ci-copilot-cli-100pct-fail-on-push(first seen 2026-06-30). Single highest-ROI reliability fix β if Smoke CI is the copilot canary, its constant red likely signals a real copilot-CLI startup regression on main. Runs: Β§29576827481, Β§29569648210.2. copilot-sdk longrun β 8Γ nonzero-token 0-turn fail (~387K tokens burned)
Eight copilot runs consumed real tokens then failed at 0 recorded turns: CLI Consistency Checker (82.7K), daily-experiment-report (68.6K), Daily Project Performance Summary (56.5K), The Daily Repository Chronicle (30.1K), Discussion Task Miner (23.3K), Smoke Copilot Sub Agents (19.3K), Copilot PR Conversation NLP Analysis (104K), Daily Max Ai Credits Test (2.9K β intentional credit test). Maps to
copilot-sdk-longrun-agentjob-fail. This is wasted spend worth root-causing.3. Avenger β 1Γ 0-tok fail (recurrence β 19)
claude/claude-haiku-4.5 0-tok/0-turn fail. Known
avenger-err-config-no-structured-logs(ERR_CONFIG "no structured log entries" on follow-up engine invocation). Long-running P1 escalation.4. codex Daily Cache Strategy Analyzer β 1Γ 0-tok fail (recurrence β 6)
codex 0-tok fail consistent with
codex-gh-aw-binary-not-found-for-mcp(MCP setup can't locate gh-aw binary). Β§29604657433.5. Metrics reporting gap β confirmed fleet-wide
metrics.Turnsandmetrics.EstimatedCostare 0 for all 177 runs (both success and failure) β so Turns=0 is a reporting artifact, not a failure signal.metrics.TokenUsageis populated inconsistently day-to-day. Knowntoken-usage-reporting-gap(recurrence β 4).β Next actions (evidence-based)
metrics.Turns/EstimatedCost/TokenUsage, or officially switch cost charts to AI-credits.copilot is the fleet's sole reliability weak point (82.7% vs pi/claude/codex β₯92%); every open failure family is copilot-engine 0-turn driver-level.
References:
Warning
Firewall blocked 1 domain
The following domain was blocked by the firewall during workflow execution:
awmgmcpgSee Network Configuration for more information.
Beta Was this translation helpful? Give feedback.
All reactions