agentsclimarketplace

Mobisys writing style

Skill brycewang-stanford/Awesome-Journal-Skills/MobiSys-Skills/skills/mobisys-writing-style

Journal-specific Claude Code/Codex skill packs covering mainstream journals — AER, QJE, Nature, Cell, 管理世界, 经济研究 & 200+ more — your fast track to getting published. | 覆盖主流期刊的 Claude Code/Codex 期刊技能包,从选题、识别策略到表格规范与审稿回复全流程,助你快速发论文。

Install
npx -y skills add brycewang-stanford/Awesome-Journal-Skills --skill mobisys-writing-style

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

Use when revising a MobiSys paper for a mobile-systems first page, the pain → system-design → on-device-evidence arc, 12-page double-column compression, double-blind wording, and claims disciplined by measured latency, energy, memory, and thermal budgets rather than superlatives, so the draft survives systems-and-services reviewers.

SKILL.md

3.6 KB, 743 tokens by cl100k_base, as published. Nobody here has run it

MobiSys Writing Style

Use this when revising the main paper. A MobiSys paper needs a compact statement of the on-device problem, the system that solves it, and the measurements that prove it — in the 12-page double-column body.

Revision rules

  • Put the mobile-system contribution on the first page: the on-device pain, why current systems fail on the device, the mechanism, and the measured payoff.
  • Name the device budget explicitly. MobiSys reviewers are sensitive to hidden compute, energy, latency, memory, and thermal assumptions.
  • Pair every major claim with a measurement on real hardware, a design argument, or a reproducibility detail — never a superlative.
  • Use the 12-page body for the system's spine; move long protocols, extra device sweeps, and extended ablations to the appendix without making the main paper unintelligible.
  • Avoid overclaiming wins when the distribution is unreported or the gain is inside run-to-run spread.
  • Maintain double-blind style in self-citations, prior-system references, acknowledgements, funding, screenshots, and artifact descriptions.

System-presentation discipline

  • State the operating point: the device, the workload, and the budget the system respects. A latency number without its device and workload is not interpretable.
  • Give each design decision a why-on-the-device justification; MobiSys readers audit whether a choice is forced by the mobile constraint or merely convenient.
  • Report costs as distributions — p50/p95 latency, energy-per-operation, memory footprint — not single "up to N×" peaks.
  • Separate measured from estimated: an energy figure from a power monitor and an energy figure from a model are different claims and must be labeled so.
  • Keep the accuracy/quality trade-off honest: a system that is faster or cooler at a cost must state the cost in the same breath.

Sentence-level rewrites

Draft patternMobiSys-safe rewrite
"Our system significantly outperforms...""cuts p95 latency from X ms to Y ms on <device> over N runs"
"Runs efficiently on mobile devices...""holds 30 fps within a 3 W budget on <device>, App. A"
"Saves substantial energy...""reduces energy-per-frame by X mJ (p50), measured on <monitor>"
"Achieves real-time performance..."Claim scoped to the device, workload, and thermal state tested

Vignette: compressing into twelve double-column pages

A draft with a runtime design, five device studies, and a sprawling background section: keep the design, one decision-critical figure (the latency-under-load curve) and one budget figure (energy-per-frame), and the accuracy-cost table; compress background into the specific failures current systems show on the device; move three device sweeps and the full protocol to the appendix with forward references. The test of a good cut: a systems reviewer should reconstruct the whole system and its evidence without opening the supplement.

Output format

[Writing diagnosis] clear / under-measured / overclaimed / overloaded
[First-page fix] <new mobile-system framing>
[Claim discipline] <claim -> measurement/design/limitation>
[Compression cuts] <move/delete/merge>
[Anonymity edits] <phrases, screenshots, or repo strings to rewrite>

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.