EPS Dashboard

Overview / Motivation

Auto-filed by the workflow-fix-on-bug protocol from a prose workflow-fix follow-up raised on task #1494 (emitting agent: code-reviewer, round 1 Minor, mechanizable: yes; parked under the recursion guard, routed by the 2026-07-18 /daily Step C parked-candidate sweep).

Goal

The implementer Gate-scope verification duty (implementer.md After-implementation step 1 + the L228 report template; the /issue SKILL.md Step 4b gate-scope duty) requires the implementer to emit the raw grep -rln basename-hit FILE LIST in its report, not just a summary.

Workflow gap

Bug observed: the #1494 round-1 implementer report omitted 7 basename-hit files (conftest + 6 test files) that the code-reviewer had to discharge itself.
Why it is a workflow gap: the Gate-scope check report template (pin-sweep: <fragments grepped> → <hits>) does not mandate the RAW hit file list, so an implementer can summarize/undercount hits and the reviewer silently inherits the discharge work — the duty exists but is not mechanically auditable from the report.
Confidence (emitter): low
verified-at-filing: grep -n 'pin-sweep' .claude/agents/implementer.md → 2 hits (L155 duty prose, L228 report template pin-sweep: <fragments grepped> → <hits>) — presence-hit context READ: the duty + template exist but neither mandates emitting the raw grep -rln hit FILE LIST (the candidate tightens, does not duplicate, the landed #1288/#1305 duty via eb4da66c68); grep -n 'Gate-scope' .claude/skills/issue/SKILL.md checked for the Step 4b mirror of the duty (2026-07-19)

Proposed change (candidate diff sketch — refine in planning)

(none — synthesized from prose follow-up)

Sketch for the planner: in implementer.md step 1(a) + the L228 report template, change pin-sweep: <fragments grepped> → <hits> to require the verbatim hit file list (e.g. pin-sweep: <fragments> → <N> hits: <file list, one per fragment>), and mirror the same requirement wherever SKILL.md Step 4b states the gate-scope duty.

Scope / surfaces

Primary target: .claude/agents/implementer.md, .claude/skills/issue/SKILL.md
Grep the workflow surface for the pattern before editing (grep -rln 'pin-sweep' .claude/ CLAUDE.md scripts/) — the experiment-implementer.md twin may carry the same duty; list every hit in the plan.

Constraints / invariants

Workflow-surface only — never experiment code, configs/, or tasks/.
scripts/workflow_lint.py --check-asks passes; any SKILL.md prose-pin tests on the edited spans stay green.
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, § Recursion guard).

Provenance

workflow_fix_target: .claude/agents/implementer.md, .claude/skills/issue/SKILL.md
fingerprint: 17973717c4c1

Verbatim surfaced prose (task #1494 events.jsonl, 2026-07-18T08:41:53Z): "source: prose-followup (code-reviewer round 1 Minor, mechanizable: yes). target_file: .claude/agents/implementer.md, .claude/skills/issue/SKILL.md (Step 4b gate-scope duty). proposed_change: the Gate-scope verification duty should require the implementer to emit the raw grep -rln basename-hit file list in its report (the round-1 report omitted 7 hit files — conftest + 6 test files — that the reviewer had to discharge itself). confidence: low. related_task: #1494."