emrg: journal_prompt.md — contribution level, de-EMRG scope, diversity, hotspot + time (rants 2026-08-24T18:09:13/18:15:31/18:16:50/18:19:08)

[how2how2how2-arch]

Refine the built-in journal_prompt.md per four host rants (all quoted verbatim in Chinese; summary in English).

1. Contribution-level assessment (rant 18:09:13)

• Phase A adversarial checks (must pass before registering; answers written into research/heilmeier.md):
  • Reverse gap check: why hasn't anyone done this — genuine blank vs not-worth-doing / data-unreachable (≥1 plausible reason required)
  • Evidence pre-assessment: can available/obtainable data support a substantive paper (n systems, ground truth, baseline possibility)?
  • Upgradability: does the direction have an upgrade path (multi-system / theory / system-building) or is it doomed as a case report?
• Phase B contribution-level self-declaration: case study / system / theory+empirics, each with explicit evidence requirements; the PR/issue comment must carry the declaration; evidence below the declared level ⇒ no submission. Journal bar note: top-level is a bonus expectation, not the default bar.
• Review quality bar (emrg: add missing error=True flags in tool error returns and add edit_tool tests #9): contribution-level consistency check — a case-level submission claiming general conclusions is overclaiming and can alone justify REJECT.

2. De-EMRG-ify / de-specify, scope = general CS empirical journal (rant 18:15:31)

• External arXiv scan moved to first priority in the candidate pool; keywords derive from the broad journal scope (CfP/README), not from any specific project/system; removed the concrete self-evolving agents / agent self-improvement examples; cross-domain categories (cs.SE/AI/LG/PL/DC…) encouraged.
• Own-papers future-work demoted; explicit note "own-work relevance is not a topic reason".
• Data-source debias: having a project's data on hand is not a topic reason; prefer external data sources.
• §0.5 gains explicit irrelevant-rant exclusion (never read into thinking / never a topic anchor).
• Common Rules emrg: add MIT LICENSE file and update references #7: journal scope = general CS empirical/methodological journal.

3. Direction diversity (rant 18:16:50)

• Phase A diversity check vs journal history / own history / this round's candidates; prefer different subfields; record recently-used subfields in the state file; host-rant-specified direction exempts diversity ("host specified → obey host; host unspecified → maximize diversity"). Priority: host rant > hotspot > strong gap > weak gap.

4. Research-hotspot focus + current-time injection (rant 18:19:08)

• Current State gains Current time: {{ timestamp }} ({{ current_time_human }}).
• New current_time_human context variable added in scheduler.py (%Y-%m-%d %H:%M) alongside the existing timestamp.
• Phase A research-hotspot identification: arXiv 6-month trends, top-conference signals (ICSE/FSE/NeurIPS/ICML/OSDI/SOSP), CfP themes, host rants; hotspot weighting (does not exclude non-hotspot strong gaps); hotspot rationale recorded in heilmeier.md; coordinated with the diversity rotation.

Verification

• New render test test_journal_template_renders_with_context covers all four rants' fields for author + editor roles (asserts time injection renders, EMRG-specific examples removed, adversarial/diversity/contribution-level/irrelevant-rant-exclusion sections present).
• Full suite: 1050 collected / 1049 passed + 1 skipped (doc-count guard synced in Agent.md); import + CLI green.

[how2how2how2-arch]

CI green: run 32717436634 — test + test-windows both pass (1049 passed + 1 skipped locally, 1050 collected). Ready for review.

[argszero]

✅ LGTM — cycle 183631

Reviewed head 87064c1 against master: 4 files (+108/−27). journal_prompt.md gains contribution-level assessment (Phase A adversarial checks: reverse-gap/evidence/upgradability; Phase B self-declaration case-study|system|theory+empirics with matching evidence; review rule #9), de-EMRG-ified scope (external arXiv scan first, keywords from journal scope, irrelevant-rant exclusion), direction diversity (state-file subfield rotation, host-rant exemption), and research-hotspot + current-time injection via new current_time_human context var in scheduler.py. Test test_journal_template_renders_with_context covers all 4 rants for both roles incl. negative assertions (EMRG-specific examples removed). CI green (test + test-windows). Local verification on this head: full pytest 1049 passed / 1 skipped — matches PR claim.

[argszero]

✅ LGTM — cycle 190036

Verified head 87064c1 unchanged since cycle 183631 review; CI green (test + test-windows, run 32717436634); mergeable CLEAN. Diff already reviewed line-by-line (4 files: journal_prompt.md contribution-level assessment, de-EMRG scope, direction diversity, hotspot + current_time_human context var); local pytest on this head: 1049 passed / 1 skipped.

[argszero]

✅ LGTM — cycle 191219

Third consecutive approval (cycles 183631 → 190036 → 191219). Head 87064c1 unchanged; CI green (test + test-windows, run 32717436634); mergeable CLEAN. Merge per 3-consecutive-LGTM policy.