EPS Dashboard

Overview / Motivation

Auto-filed by the /daily 2026-07-28 problem sweep (transcript miners over the day's 55 session transcripts). Source: group-H P4.

Pre-split known multi-deliverable implementer builds at dispatch instead of recovering with the same split after a context-ceiling death.

Bug observed: #1775's implementer was dispatched with a 7-script deliverable set; it died at the context ceiling after 139 tool calls / ~63 min (bulk of the work had landed durably; ~15 min triage + respawn overhead). The recovery — sequential micro-scoped rounds — is exactly the split available at dispatch time from the plan's own deliverable count.
Why it is a workflow gap: the micro-scoped split exists only as a POST-death respawn recipe (SKILL.md ~2845); nothing keys it proactively off the plan's declared deliverable count.
Confidence (emitter): medium (inferred; the death + recovery are transcript-verified)
verified-at-filing: grep -n 'micro-scop' .claude/skills/issue/SKILL.md → respawn recipe at ~2845-2848 (recovery-side only) (2026-07-29 UTC).

One dispatch-step clause: deliverable count > threshold ⇒ sequential micro-scoped rounds by default; cite the #1090/#1775 precedents.

Workflow-surface only — never experiment code, configs/, or tasks/.
scripts/workflow_lint.py --check-asks passes; ruff on touched files passes.
The spawned session runs under a workflow_fix_target: Provenance line — recursion guard applies (it parks, never auto-routes, its own subagents' workflow-fix candidates).