workflow-fix: strip v4 Methodology details in body audit

Overview / Motivation

Auto-filed by the /daily Step-C parked-candidate sweep (2026-07-08) from a candidate parked on task #969 (recursion-guarded workflow-fix session).

Goal

Stop v4 clean-result bodies from false-positiving the body-discipline audit on verbatim sample-data rows that live in ## Methodology by spec.

Workflow gap

• Bug observed: strip_data_example_blocks (audit_clean_results_body_discipline.py:505) drops
  example blocks ONLY inside the v3 ## Data section (re-enters only on ## Data, :539-540) — v4 ## Methodology

  sample-data blocks are never stripped, so verbatim example rows carrying condition codes / interval forms can false-positive audit categories on v4 bodies.
• Why it is a workflow gap: the fix targets the workflow surface (scripts/audit_clean_results_body_discipline.py); the originating session was recursion-guarded and could not route it.
• Confidence (emitter): see parked note below.

Proposed change (candidate diff sketch — refine in planning)

- # re-enter only when the new H2 is `## Data` itself
+ # re-enter when the new H2 is `## Data` (v3) or `## Methodology` (v4 sample-data slot)
- in_section = line.strip() == "## Data"
+ in_section = line.strip() in ("## Data", "## Methodology")

Scope / surfaces

• Primary target: scripts/audit_clean_results_body_discipline.py
• Grep the workflow surface for the pattern before editing 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.
• The spawned session runs under EPM_WORKFLOW_FIX_SESSION=1 / a workflow_fix_target: Provenance line — it MUST NOT auto-route its own subagents' workflow-fix candidates (recursion guard).

Provenance

• workflow_fix_target: scripts/audit_clean_results_body_discipline.py
• origin: parked candidate on task #969 at 2026-07-04T07:45:46Z

Verbatim parked note:

routed: parked — running under workflow_fix_target Provenance (recursion guard, .claude/rules/workflow-fix-on-bug.md § Recursion guard); NOT auto-filed. source: prose-followup (planner assumption-12 observation, #969 Phase 1). target_file: scripts/audit_clean_results_body_discipline.py. bug_observed: strip_data_example_blocks gates on '## Data' only (line ~496), so v4 '## Methodology'

sample-data blocks are NOT stripped for ANY audit category — verbatim example rows carrying condition codes / interval forms can false-positive on v4 bodies. proposed_change: extend strip_data_example_blocks (or add a v4 branch) to also strip

blocks under v4 ## Methodology. confidence: medium. related_task: #969.

Merged sibling candidate (s4-audit-v4-sample-block-exempt, from task:970 at 2026-07-04T07:53:33Z)

• bug_observed: strip_data_example_blocks exempts
  example blocks only inside the v3 ## Data section; v4 bodies carry spec-mandated verbatim sample blocks under ## Methodology -> **Sample training/evaluation data + completions:**, which therefore remain in the scan source for EVERY non-exempt category — verbatim rows cannot be reworded, so a caps token in an unfenced v4 sample block can false-FAIL the audit.
• proposed_change: Extend the
  exemption to the v4 Methodology sample slot for ALL scan categories (mirror the existing ## Data mechanism), with a v4-details-stays-clean test plus a #763-Results-still-flags test.
• origin note (verbatim): NOT fixed: scripts/audit_clean_results_body_discipline.py strip_data_example_blocks (l.505-552) still enters
  drop-mode ONLY inside the v3 ## Data section ('Any H2 ends a ## Data block... re-enter only when the new H2 is ## Data itself'); no ## Methodology / v4 sample-slot handling anywhere. The completed audit-script fixes since (#1015 interval_inline bound-form; #969; #731; #892) are unrelated. Distinct (target_file, fingerprint 770f63a8859c) from #970's own fix. No retraction in #970 events.