Problem
On 2026-06-12, issues #38793 and #38794 were filed with identical titles and bodies, 2 seconds apart, by an automated agent workflow. This is a clear race-condition / missing-deduplication pattern in the Workflow Health Monitor or Failure Investigator agent.
Evidence
| Issue |
Title |
Created |
| #38793 |
[aw] Code Simplifier: 4-day persistent failure streak — needs root-cause fix (P1) |
2026-06-12T06:17:39Z |
| #38794 |
[aw] Code Simplifier: 4-day persistent failure streak — needs root-cause fix (P1) |
2026-06-12T06:17:41Z |
Both issues contain identical bodies. Prior runs also noted this risk when #aw_cs4d temporary ID was introduced but deduplication was not enforced.
Root Cause
Agent workflow that creates tracker issues does not check for pre-existing open issues with the same title before filing. When the temporary_id pattern (#aw_cs4d) was used internally but not enforced via a title-search guard, two parallel invocations (or a retry) produced duplicates.
Impact
- Noise: Duplicate issues consume safe-output quota (2× per event)
- Confusion: Human reviewers see the same issue twice in triage
- Debt: One duplicate must be manually closed
- Pattern risk: Any agent that creates tracker issues is vulnerable if it runs more than once without a dedup check
Recommended Fix
- Before
create_issue: run gh issue list --search "TITLE" --state open and skip if a match exists.
- Use
temporary_id to cross-reference in the same run; confirm the ID is stable and surfaced in safe-output validation.
- Add dedup logic to Workflow Health Monitor and Failure Investigator prompts explicitly.
- Consider a shared
shared-alerts.md lock — the file already lists "DO NOT RE-FILE" entries; agents should verify this list before creating any tracker issue.
Effort Estimate
- Prompt update for affected workflows: ~1–2 hours
- Expected improvement: Reduce duplicate issue rate by ~95%
References
- §27419589668 — Agent Performance Analyzer run that detected this pattern
- Agent-performance-latest.md:
shared-alerts.md already documents the "Do Not Re-File" pattern but agents are not reliably checking it
Generated by ⚡ Agent Performance Analyzer - Meta-Orchestrator · 501 AIC · ⌖ 20.6 AIC · ⊞ 22.2K · ◷
Problem
On 2026-06-12, issues #38793 and #38794 were filed with identical titles and bodies, 2 seconds apart, by an automated agent workflow. This is a clear race-condition / missing-deduplication pattern in the Workflow Health Monitor or Failure Investigator agent.
Evidence
Both issues contain identical bodies. Prior runs also noted this risk when
#aw_cs4dtemporary ID was introduced but deduplication was not enforced.Root Cause
Agent workflow that creates tracker issues does not check for pre-existing open issues with the same title before filing. When the
temporary_idpattern (#aw_cs4d) was used internally but not enforced via a title-search guard, two parallel invocations (or a retry) produced duplicates.Impact
Recommended Fix
create_issue: rungh issue list --search "TITLE" --state openand skip if a match exists.temporary_idto cross-reference in the same run; confirm the ID is stable and surfaced in safe-output validation.shared-alerts.mdlock — the file already lists "DO NOT RE-FILE" entries; agents should verify this list before creating any tracker issue.Effort Estimate
References
shared-alerts.mdalready documents the "Do Not Re-File" pattern but agents are not reliably checking it