Fix the safe-outputs update_issue handler: a target:"triggering" resolution miss on a scheduled run must be a soft-skip, not a job-failing ##[error]. LintMonster's agent succeeded and created issue #39481 + discussion #39482, yet the safe_outputs job went red and auto-filed nothing actionable. This is a false-failure, not an outage.

Affected workflow & run

• Workflow: LintMonster (.github/workflows/lint-monster.lock.yml), schedule event
• Failed run: §27592639107 — safe_outputs job, step "Process Safe Outputs"
• Agent job: ✅ success (6.1m, 44 turns) — create_issue [lint-monster] Optimize Set Usage: map[string]bool → map[string]struct{} (187 findings) #39481 and create_discussion [lint-monster] 🧌 LintMonster Daily Report — 2026-06-16 #39482 both completed
• Regression window: 06-13/06-14/06-15 all ✅ succeeded; first failure 06-16 (fresh regression, not a standing outage)

Root cause

The agent emitted 3 update_issue messages with target:"triggering" while running under a scheduled trigger that has no issue context. The handler logs a warning ("skipping update_issue") but then also records each as a failed message and core.setFaileds the job:

Processing message 1/6: update_issue
##[warning]Target is "triggering" but not running in issue context, skipping update_issue
##[error]✗ Message 1 (update_issue) failed: Target is "triggering" but not running in issue context, skipping update_issue
...
=== Processing Summary === Total: 6  Successful: 2  Failed: 3  Cancelled (code push failed): 1
##[error]3 safe output(s) failed:

The wording is self-contradictory: "skipping" implies a no-op, but the same outcome is counted as a hard failure. create_issue #39481 and create_discussion #39482 succeeded in the same run.

audit-diff vs nearest successful baseline (§27522679098): zero firewall anomalies, posture read_only unchanged, 0 blocked requests. This is not infra/MCP/network drift — it is purely safe-outputs message handling.

Secondary contributing signals (same run)

• GH_AW_DETECTION_REASON=parse_error, GH_AW_DETECTION_CONCLUSION=warning → threat-detection warn policy (WTD3) aborted the assign_to_agent message.
• "1 message(s) were cancelled because a code push operation failed" — a push_to_pull_request_branch/code-push message was cancelled.
  These are independent of the primary update_issue escalation but worth noting; the 3 update_issue errors are what turned the job red.

Probable fix (pick one or both)

1. Handler: When target:"triggering" cannot resolve to an issue in a scheduled/non-issue context, treat update_issue as a skipped/no-op (warning only) and exclude it from the failed-message count, instead of core.setFailed. Align the log with the behavior.
2. Workflow config: If LintMonster legitimately needs to update a known issue on schedule, set update-issue: target: '*' and have the agent pass an explicit issue_number; otherwise remove the update_issue capability from a schedule-only report workflow so the agent stops emitting it.

Success criteria / verification

• A scheduled LintMonster run that emits update_issue with unresolvable target:"triggering" ends with safe_outputs green when all other messages succeed.
• The "skipping update_issue" path no longer increments the failed-message counter or calls core.setFailed.
• Next scheduled LintMonster run (03:44 UTC) completes with conclusion success.

Evidence run IDs

• Failed: 27592639107 · Successful baseline: 27522679098 (06-15), 27487423958 (06-14), 27455558682 (06-13)
  Related to Workflow Health Manager - Meta-Orchestrator - Issue Group #29109

Generated by 🔍 [aw] Failure Investigator (6h) · 263.8 AIC · ⌖ 12.1 AIC · ⊞ 4.5K · ◷

• expires on Jun 23, 2026, 12:52 AM UTC-08:00