EPS
← All tasks·#1305Completed

daily-fix: implementer pre-report 9c touched-scope run

kind: infra#wf-fix#wf-fix-fp:c2850d74e9f8#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 2026-07-13 from the transcript problem sweep (session 5cbde72b, task #1288 — the only Step 9c gate FAIL round of the day's 15-task infra drain).

Goal

Make the implementer's pre-report verification scope match the Step 9c gate's touched-scope selection, so a diff that changes a pinned literal cannot pass local verification and then fail the gate.

Workflow gap

  • Bug observed: #1288's implementer round 1 changed a pinned literal in the Step 10d command sequence but did not update the 3 tests pinning it in tests/test_step10d_guard3.py; its self-chosen local verification (14 pin/regression tests, all green) missed them, while the 9c touched-scope selection (37 files) ran them → gate round-1 FAIL, one full implementer + reviewer + gate rework round (~30 min).
  • Why it is a workflow gap: the implementer spec never tells the implementer to run (or approximate) the same select_step9c_tests.py --base main selection the gate will run — the verification-scope mismatch is structural, not a one-off oversight.
  • Confidence (emitter): medium-high (single clean incident, structural cause)
  • verified-at-filing: grep -n "select_step9c_tests" .claude/agents/implementer.md .claude/agents/experiment-implementer.md → 0 hits in both (2026-07-14 UTC) — the pre-report selection duty is absent from both implementer specs.

Proposed change (candidate diff sketch — refine in planning)

In .claude/agents/implementer.md (and the Step 4b brief template in .claude/skills/issue/SKILL.md, plus experiment-implementer.md if the planner finds the same gap there):

+ Before reporting success, run the SAME touched-scope test selection the
+ Step 9c gate will run: `uv run python scripts/select_step9c_tests.py --base main`
+ from the worktree, and run the selected files — or at minimum every test
+ file that greps any literal/symbol the diff changed. A local verification
+ scope narrower than the gate's selection is the #1288 rework shape.

Scope / surfaces

  • Primary targets: .claude/agents/implementer.md, .claude/skills/issue/SKILL.md (Step 4b brief)
  • Secondary: .claude/agents/experiment-implementer.md

Constraints / invariants

  • Workflow-surface only; scripts/workflow_lint.py default run passes.
  • Recursion guard applies to the spawned session.

Provenance

  • workflow_fix_target: .claude/agents/implementer.md, .claude/skills/issue/SKILL.md
  • fingerprint: c2850d74e9f8

Origin: transcript-mined (session 5cbde72b, ~08:00-08:30Z; task #1288 events epm:test-verdict round-1 FAIL → round-2 PASS). Not a parked candidate — surfaced by the /daily problem sweep.

Activity