Mathematics Paper Campaign
SkillDocs & knowledgeRun a careful, multi-step campaign on a complete mathematical LaTeX paper, either building it from a proof outline, notes, a formal (Lean) development, or an earlier draft, or taking an already checked manuscript through phased polishing (reduce, cite, linearise, rewrite, introduction, final audit) to a publication-ready paper. Use for whole-paper writing, rewriting, or "make this paper better" requests that need statement-by-statement control, independent reviews, ledgers, and rendered-page checks.
Use Mathematics Paper Campaign in Claude, ChatGPT or Ahel Desktop
Free. Sign in, add Mathematics Paper Campaign and connect your AI. About a minute.
Also: Claude Code · Cursor · Codex
Then ask your AI: use the Mathematics Paper Campaign skill
Details
Instructions available. Your AI can read the instructions. Execution depends on the setup they require.
Account requirements not reviewed. Check the skill instructions before use; Ahel provides instructions and does not run this skill.
No other account needed.
Add Ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.
What this skill tells your AI
The instructions your AI receives, as published by amelieloher/math-paper-skills in skills/math-paper-campaign/SKILL.md and read by Ahel’s review.
Own the mathematical meaning, the public statement surface, the architecture, source fidelity, the canonical manuscript, and the acceptance gates of a whole paper. Use the latex-math-writing skill for the prose and typesetting of every packet; this skill controls the process around it.
The governing rule: work on one mathematical statement at a time. Never turn an outline into a section, or a messy section into a clean one, in a single undifferentiated pass.
Carry the campaign through successive packets until the requested paper is complete. Do not stop after producing plans or review notes when the next packet is ready. Pause only for a genuine mathematical gap, a consequential author decision, missing source material, or authority the user has not granted.
Choose the route
| Starting point | Route |
|---|---|
| a proof outline, notes, a formal development, or a draft whose proofs are not yet all written and checked | Route A: build — establish and write a correct, reviewed paper |
| a manuscript whose statements and proofs have been checked at a named revision, to be made clear and publication-ready | Route B: polish — Phases 0–6: freeze, reduce, cite, linearise, rewrite, introduction, final audit |
| an old paper to be restructured around a new spine | Route A's rewrite mapping, then Route B once the mathematics is checked |
A paper built under Route A that needs substantial compression or restructuring continues into Route B from Phase 0; a paper built carefully needs only Route B's introduction phase and final audit (Phases 5 and 6). If only part of a manuscript is checked, run Route B on the checked packets and Route A on the rest, keeping the unchecked material blocked and visible.
A request for one phase or one local edit does not authorise restarting the whole campaign. A narrow edit belongs to latex-math-writing alone.
What both routes share
Shared controls defines what every campaign uses:
- one canonical manuscript and one canonical editor;
- bounded statement packets, at most one load-bearing packet under edit at a time;
- a compact ledger with separate mathematical and manuscript statuses;
- the fail-closed drift protocol for any change to a frozen statement;
- independent, read-only reviewers with fresh context;
- the final publication audit and completion test.
Style
Use the author's or venue's designated style. When none is named, use the house style of latex-math-writing: standard article layout, Computer Modern text and mathematics, dark-blue navigation links, mechanism-first explanations. Record the style authority in the brief and include it in writing and exposition-review packets; do not reopen a settled style choice. Current instructions, mathematical contracts, venue requirements, and other authors' style authority keep their priority.
Prohibited shortcuts
- Do not ask an agent to "fill in" or "clean up" an entire section in one pass.
- Do not write around an unresolved mathematical gap, or polish it away.
- Do not promote a diagnostic note or formal API lemma directly into the paper; do not add paper lemmas for future formalisation convenience.
- Do not use the proof author as the only referee of a load-bearing proof.
- Do not keep competing proofs or manuscripts active without an explicit reason.
- Do not accept "compiles" as evidence of mathematical or typographic completion.
- Do not write the final introduction while the principal theorem surface is still moving.
- Do not let control vocabulary (packet, ledger, gate, consumer) reach the manuscript.
Related skills
latex-math-writing— prose, notation, displays, and rendered-page verification inside each packetvisual-paper— an interactive companion website built from the finished paper- For formalising the same mathematics in Lean, see LeanAutoformalizationSkills
Signals
- GitHub stars
- 27
- Forks
- 2
- Last commit
- Oct 2026
Advanced
- Item type
- skill
- Key
math-paper-campaign- Source
- github.com/amelieloher/math-paper-skills
github.com/amelieloher/math-paper-skills
Related picks
Skill · k-dense-ai
The pick for LaTeXlatex-drawing-guide
Skill · brycewang-stanford
The pick for LaTeXhandoff
Skill · mattpocock
More in Docs & knowledgecanvas-design
Skill · anthropics
More in Docs & knowledgedoc-coauthoring
Skill · anthropics
More in Docs & knowledgepopups
Skill · coreyhaines31
More in Docs & knowledge