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
{{ message }}
Repository navigation
Fleet Monitor reports legacy per-role stub drift as agent-ingress.yml drift (alerts and remediation plan target repos that never collapsed) #2289
For a repo without an ingress, classify_role_drift correctly falls back to comparing the legacy per-role file (for example .github/workflows/dev-lead.yml) with the legacy canonical (standards/workflows/dev-lead.yml).
But stub_drift_alert_json stamps every DRIFTED row with the registry row's stub_path, which is .github/workflows/agent-ingress.yml. So [Fleet Monitor] Dev-lead stub drift #2284–[Fleet Monitor] Ci-failure-analyst stub drift #2288 tell dev-lead to "re-sync" a .github/workflows/agent-ingress.yml stub in .github, .github-private, ContentTwin, TalkTerm, bmad-bgreat-suite and google-app-scripts. None of them has, or should have, an ingress. Following that instruction is exactly the double-dispatch trap ADR-0007 guards against.
The dry-run remediation plan likewise lists agent-ingress.yml writes for those repos. Live mode would fail them, because _remediate_role_block errors when the repo has no ingress file. The plan and the alert are still wrong.
The five issues #2284–#2288 are on dev-lead:hands-off until this is fixed.
Fix (minimal)
In scripts/fleet_monitor.sh, exclude role rows whose classification came from the legacy fallback (repo has no agent-ingress.yml) from fleet_stub_drift.json (alerts and remediation input). Keep them in the per-stub report table, marked as legacy, so the information is not lost.
Rows from a repo that has an agent-ingress.yml (block comparison) keep alerting and remediating as today.
Do not change the whole-file initiative-planner / initiative-driver rows, classify_role_drift, or remediation's live guards.
Bats coverage:
(a) a legacy repo whose per-role file drifts from the legacy canonical appears in the report but not in fleet_stub_drift.json;
(b) a collapsed repo whose block drifts is in fleet_stub_drift.json with stub_file = .github/workflows/agent-ingress.yml;
(c) the remediation input never lists agent-ingress.yml for a repo that has no ingress.
Acceptance
shellcheck and bats are green.
After merge, a dry-run Fleet Monitor shows the collapsed-role drift issues listing only repos that actually carry an agent-ingress.yml (today only petry-projects/markets).
Tag role rows classified via the legacy fallback (repo has no agent-ingress.yml) and keep them out of fleet_stub_drift.json (alerts + remediation input). They stay in the per-stub report, marked as legacy.
Implementation Checklist
scripts/fleet_stub_drift.sh — stub_drift_row_role appends a 5th TSV field legacy when the row came from the legacy fallback. Ingress-classified rows and whole-file rows keep the 4-field shape.
The implementation for issue #2289 is durable: commits are pushed and a pull request is open. This record is backed by verifiable artifacts you can confirm in one click.
Review happens on the PR. Acceptance criteria and test results are evidenced by the PR's diff and CI checks — not asserted here. If a later stage fails, this claim is retracted in place.
Activity
don-petry commentedon Oct 11, 2026
Dev-Lead Implementation Plan
Issue: #2289 — Fleet Monitor reports legacy per-role stub drift as agent-ingress.yml drift
Scope
Tag role rows classified via the legacy fallback (repo has no
agent-ingress.yml) and keep them out offleet_stub_drift.json(alerts + remediation input). They stay in the per-stub report, marked as legacy.Implementation Checklist
scripts/fleet_stub_drift.sh—stub_drift_row_roleappends a 5th TSV fieldlegacywhen the row came from the legacy fallback. Ingress-classified rows and whole-file rows keep the 4-field shape.scripts/fleet_stub_drift.sh—stub_drift_alert_jsondrops rows taggedlegacy.scripts/fleet_stub_drift.sh—generate_stub_drift_reportmarks legacy DRIFTED rows in the table ("legacy per-role stub — not alerted").classify_role_drift, the initiative-planner/driver rows, or remediation's live guards.Tests to Write
tests/fleet_stub_registry.bats: a legacy repo whose per-role file drifts shows in the report (marked legacy) but not infleet_stub_drift.jsonfleet_stub_drift.jsonwithstub_file=.github/workflows/agent-ingress.ymland its rolebuild_remediation_planon the emitted JSON listsagent-ingress.ymlonly for the repo that has onetests/fleet_stub_drift.batsfor the legacy tag and the alert filterTest Command
bats tests/andbash .dev-lead/scripts/dev-lead-lint.shdon-petry commentedon Oct 11, 2026
Dev-Lead: Implementation Complete — PR #2292
The implementation for issue #2289 is durable: commits are pushed and a pull request is open. This record is backed by verifiable artifacts you can confirm in one click.
afc3042d3608097966ec7f7fbeb4facf8434bf7dmain: 4 file(s), +125/-12 linesReview happens on the PR. Acceptance criteria and test results are evidenced by the PR's diff and CI checks — not asserted here. If a later stage fails, this claim is retracted in place.
feat: implement issue #2289 — Fleet Monitor reports legacy per-role s…