EPS
← All tasks·#1677Completed

daily-fix: label unverified claims in route-2 bodies

kind: infra#wf-fix#wf-fix-fp:f15904190f8c#daily-auto-filed
track:

Highlight text on any card (body, plan, or an event) to anchor a comment, or leave a whole-task comment here. Mention @claude to summon a reply.

No comments yet.

Overview / Motivation

Auto-filed by /daily 2026-07-24 problem sweep (sessions 1a210c3c/6dbde886/6d2780e7/d92dbcd2, tasks #1655/#1662/#1646/#1660).

Goal

Filed bodies should carry honest epistemic labels on claims the filer could not mechanically verify, so spawned sessions treat them as hypotheses.

Workflow gap

  • Bug observed: 4 of 6 sessions in the 07-24 infra wave had to correct filed-body premises: #1655 (fact-check: "commit b66910d748 shows the account list was intact" — filed root-cause wrong), #1662 (fact-checker: bare-form terminate DOES sweep suffixed pods — filed assumption wrong), #1646 (filed "~200s alone" premise refuted by measurement), #1660 (filed "3600s" a mis-recollection vs train_behavior_fullft.py:650). Counted from each session's pipeline-record messages. Each burned a correction round pre-approval.
  • Why it is a workflow gap: the verified-at-filing mandate covers grep-verifiable claims; timings, incident mechanisms, and API-behavior premises are asserted as fact with no labeling convention, so planners inherit them as premises.
  • Confidence (emitter): high.
  • verified-at-filing: n/a — behavioral observation from the four sessions' pipeline records (task events #1655/#1662/#1646/#1660, 2026-07-24); the SKILL.md route-2 section (read 2026-07-25) has no hypothesis-labeling clause (absence bind, context read).

Proposed change (candidate diff sketch — refine in planning)

In .claude/skills/daily/SKILL.md route 2 (and the wf-fix body template note): claims not verified mechanically at compose time — timings, incident root-causes, API behavior — are written as unverified hypothesis — verify at plan time: <claim>, never as bare fact. One sentence + example in the template.

Scope / surfaces

  • Primary target: .claude/skills/daily/SKILL.md (route-2 composition; optionally .claude/rules/workflow-fix-on-bug.md body template)

Constraints / invariants

  • Workflow-surface only — never experiment code, configs/, or tasks/.
  • scripts/workflow_lint.py no-flags run passes; ruff on touched files passes.
  • This session runs under a workflow_fix_target: Provenance line — recursion guard applies.

Provenance

  • fingerprint: f15904190f8c

  • workflow_fix_target: .claude/skills/daily/SKILL.md

Activity