workflow-fix: corpus WARN on H1 vs frontmatter title drift

Overview / Motivation

Auto-filed by the /daily Step-C parked-candidate sweep (2026-07-08, slice 5) from a candidate parked on task #1110 by a recursion-guarded workflow-fix session.

Goal

Add a corpus-wide WARN-level H1-vs-frontmatter-title sync check to the clean-results body-discipline auditor, so post-gate title drift on parked/grandfathered bodies gains a mechanical surface.

Workflow gap

• Bug observed: The #1110 check binds only at gate-time verify_task_body.py runs; a retitle AFTER the final gate (on awaiting_promotion or completed bodies) is never re-verified and the corpus auditor does not compare H1 vs frontmatter title.
• Why it is a workflow gap: Post-gate title drift (the #432/#458 divergence shape) has no mechanical detection surface anywhere.
• Confidence (emitter): medium (formal candidate block; Claude alternatives-lens critic on the #1110 plan, seconded by the reconciler)
• Triage evidence (2026-07-08): NOT fixed on main: the gate-time H1-vs-frontmatter check landed in verify_task_body.py (commit 1dd959e650, issue-1110) but audit_clean_results_body_discipline.py has no H1-title-sync corpus check (grep for find_h1_title / 'frontmatter title' / H1: no relevant hits) — post-gate retitles on awaiting_promotion/completed bodies remain mechanically unwatched, exactly the parked ask. Completed #1015 targeted this file for a different bug (interval_inline) — not a dedup. No retraction.

Proposed change (candidate diff sketch — refine in planning)

+ def audit_h1_title_sync(body, fm): ...  # WARN row per divergent sentinelled body
+   (reuse verify_task_body.find_h1_title + whitespace-collapse normalization)
+ wire into the auditor's per-body check list; WARN only, never FAIL

Scope / surfaces

• Primary target: scripts/audit_clean_results_body_discipline.py
• Secondary: reuse (import or mirror) scripts/verify_task_body.py's find_h1_title/normalization; add fixtures to the auditor's test file.
• Grep the workflow surface for the pattern before editing (grep -rln '<pattern>' .claude/ CLAUDE.md scripts/) and update every hit; list them in the plan.

Constraints / invariants

• Workflow-surface only — never experiment code, configs/, or tasks/.
• scripts/workflow_lint.py --check-asks passes; ruff on touched files passes; if workflow.yaml or CLAUDE.md change, they stay consistent with the rule file.
• This session runs under EPM_WORKFLOW_FIX_SESSION=1 and carries a workflow_fix_target: Provenance line — it MUST NOT auto-route any of its own subagents' workflow-fix candidates (recursion guard).

Provenance

• workflow_fix_target: scripts/audit_clean_results_body_discipline.py
• origin: parked candidate on task #1110 at 2026-07-07T13:04:29Z

Verbatim parked note:

parked — running under workflow_fix_target Provenance (recursion guard); LOGGED for the next orchestrator/human pass, NOT auto-filed.

target_file: scripts/audit_clean_results_body_discipline.py bug_observed: Retitles AFTER the final gate (on awaiting_promotion post-critic-PASS or completed bodies — the #432/#458 divergence shape) are never re-verified: no verifier run revisits those bodies, and the corpus auditor does not compare H1 vs frontmatter title. why_workflow_gap: The #1110 check binds only at gate-time verify runs; post-gate title drift on grandfathered/parked bodies has no mechanical surface. proposed_change: Add a corpus-wide WARN-level H1-vs-frontmatter-title comparison to audit_clean_results_body_discipline.py (reuse verify_task_body's find_h1_title + whitespace-collapse normalization; WARN only, never FAIL — post-gate remediation is a human call). diff_sketch: |

• def audit_h1_title_sync(body, fm): ... # WARN row per divergent sentinelled body
• (wire into the auditor's per-body check list) confidence: medium related_task: #1110

(Origin: Claude alternatives-lens critic concern on the #1110 plan, seconded by the reconciler as "a valid follow-up note".)