cpw-write
Plan, draft, revise, polish, or translate cryptography and security manuscripts, from a single passage to a full-paper outline.
Skill by Noi1r
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/noi1r-cpw-write
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: security, writing, research, productivity
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 26, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/cpw-write
What's inside
- Skill name
cpw-write- Description
- Plan, draft, revise, polish, or translate cryptography and security manuscripts, from a single passage to a full-paper outline. Use as the single CPW writing entrypoint; read its internal guides for the requested task. Not for toolkit maintenance or independent research execution.
- License
- MIT
- Instructions
- 120 lines in SKILL.md · view on GitHub
Files (25)
- SKILL.md
- materials-to-draft.md · examples/
- revision-decisions.md · examples/
- toy_abstracts.md · examples/
- toy_intro_and_contributions.md · examples/
- toy_overview_and_related.md · examples/
- author-style.md · references/
- collaborative-polishing.md · references/
- drafting-from-materials.md · references/
- field-language-and-clarity.md · references/
- review-comparison.md · references/
- 00_overall_structure.md · references/sections/
- 01_title.md · references/sections/
- 02_abstract.md · references/sections/
- 03_introduction.md · references/sections/
- 04_contributions.md · references/sections/
- 05_technique_overview.md · references/sections/
- 06_related_work.md · references/sections/
- 07_preliminaries.md · references/sections/
- 08_main_construction.md · references/sections/
- 09_security_correctness.md · references/sections/
- 10_experiments.md · references/sections/
- 11_conclusion.md · references/sections/
- 12_appendix.md · references/sections/
- writing-from-related-work.md · references/
SKILL.md (first lines)
# CPW manuscript writing
Use this entrypoint for all manuscript writing tasks. Load only the internal
guide needed for the author's current request and stay within the agreed scope.
These guides are reference documents for the current session; they do not
spawn separate skill calls or subagents.
## Start from the requested work
Identify the target passage or section, its job in the paper, its intended
reader, and its surrounding dependencies. Preserve any clear structure and
established terminology the author already uses. Read enough surrounding context
to see what the reader already knows entering the passage and what the passage
must establish for later sections. Reuse prior writing decisions and source
checks whenever their inputs have not changed.
Work directly from the manuscript, author-provided text, bibliography, and
supporting notes. When present, read `STATE.md` for file paths and the writing
plan, `CLAIMS.md` for exact supported claim wording, `EVIDENCE.md` for numerical
provenance, and `LITERATURE.md` for prior-work comparisons. A local edit or
polishing task can proceed directly from the manuscript and existing notes—do
not initialize project ledgers, run a fresh literature search, or build a
style profile just to revise a passage.
For a new paper or major reorganization, establish the actual technical results
first, study suitable reference papers, and propose an outline tailored to the
work. Record the plan in `STATE.md` or existing project notes. Keep technical
proximity (closest prior art) distinct from rhetorical usefulness (well-written
exemplar papers). When drafting new prose from raw notes, proofs, or benchmark
records, use [drafting from research materials](references/drafting-from-materials.md)
to build the core argument, explanation order, and complete draft text.
## Read only the guidance needed now
Resolve links relative to this skill directory, following its symlink if needed.
Read the guide matching the current task; follow cross-links only when a
concrete decision requires them. Never preload every section guide for a local edit.
| Requested work | Internal guide |
|---|---|
| Paragraph or section polishing, translation, iterative revision | [Collaborative polishing](references/collaborative-polishing.md) |
| A batch of proposed revisions for asynchronous review | [Paragraph comparison booklets](references/review-comparison.md) |
| New paper or substantial reorganization | [Writing from relevant papers](references/writing-from-related-work.md), then [overall structure](references/sections/00_overall_structure.md) |
| First manuscript draft or a new section from notes and results | [Drafting from materials](references/drafting-from-materials.md), with the relevant chapter guide below |
| Title | [Title](references/sections/01_title.md) |
| Abstract | [Abstract](references/sections/02_abstract.md) |
| Introduction or contributions | [Introduction](references/sections/03_introduction.md) and, when needed, [contributions](references/sections/04_contributions.md) |
| Technical overview | [Technical overview](references/sections/05_technique_overview.md) |
| Related-work prose | [Related work](references/sections/06_related_work.md) |
| Background and notation | [Preliminaries](references/sections/07_preliminaries.md) |
| Construction, protocol, or algorithm exposition | [Main technical content](references/sections/08_main_construction.md) |
| Presentation of security, correctness, or analysis | [Security and correctness](references/sections/09_security_correctness.md) |
| Evaluation, including analytical cost comparisons | [Evaluation](references/sections/10_experiments.md) |
| Conclusion or appendix | [Conclusion](references/sections/11_conclusion.md) or [appendix](references/sections/12_appendix.md) |
| Deriving a personal style profile, when explicitly requested | [Author style](references/author-style.md) |
When diagnosing awkward phrasing, dense nominalizations, or unclear terminology,
read [field language and clarity](references/field-language-and-clarity.md).
[Synthetic revision examples](examples/revision-decisions.md) illustrate structural
and attribution choices; treat them as decision patterns, not sentence templates.
## Shared writing decisions
- **Fix structure before surface wording.** Give every paragraph one clear job
in the argument. Resolve misplaced premises, missing logical steps, and
duplicated explanations before tuning individual words. Choose sentence and
paragraph boundaries to clarify the reasoning; shorter text is only better
when it keeps the necessary motivation, conditions, and causal links intact.
- **Evaluate flow in context (Given-before-New).** Judge a passage by what the
reader has already seen at that point in the manuscript. A body section can
rely on earlier overviews, definitions, and notation without re-teaching them;
add only the local reminders and transitions needed here. Within sentences and
paragraphs, lead with familiar context or previously introduced objects and
place the new technical observation, constraint, or result at the end. When a
deliverable must stand alone (such as an abstract), make its context self-contained.
- **Pair exact domain terms with plain action verbs.** Use the subfield's
established names consistently—never swap synonyms for a defined technical
object (`witness`, `commitment`, `oracle`, `simulator`, `distinguisher`). Around
those technical nouns, use direct verbs (`computes`, `sends`, `checks`, `bounds`,
`reduces`, `interpolates`) instead of abstract noun chains or empty intensifiers.
Integrate math cleanly into English grammar: pair new symbols with a descriptive
noun on first mention (for example, "the ring dimension $d$") and never start a
sentence with a bare symbol.
- **Never strengthen a claim through editorial polish.** Preserve all assumptions,
quantifiers, adversarial capabilities, probability spaces, security models, and
comparison regimes. Keep language improvements strictly separate from technical
corrections; fluent prose cannot supply a missing reduction step, causal link,
or novelty claim.
- **Keep evidence categories distinct.** Clearly distinguish proved theorems,
analytical cost models, manuscript derivations, local benchmark measurements,
numbers quoted from prior papers, and heuristic estimates. Cite the primary
source that establishes each fact: cite a standard for normative parameters or
definitions, and cite the specific analysis or implementation paper for its
bounds or timings.
- **Keep every number traceable.** Trace modified numerical claims to their
underlying logs, tables, or calculations. Update `EVIDENCE.md` and `\ev{...}`
links when the project uses them; otherwise rely on existing source records.
Flag any missing data or citation explicitly rather than inserting a placeholder
number or an unverified reference.
## Deliver and check
After a review or a batch of multi-passage edits, offer a
[comparison booklet](references/review-comparison.md) once if asynchronous side-by-side
review would help the author; do not make it a prerequisite for normal editing.
When revising, return the complete revised passage along with a brief summary of
substantive changes unless the author asks for line notes or discussion only.
If a passage already works well, say so and leave it untouched. Keep citation
keys, `\label`/`\ref` targets, LaTeX macros, and revision-tracking markup intact
unless authorized to change them. Discussing candidate wording in chat does not
authorize editing manuscript files on disk; edit source files only when requested.
When the author changes a section's framing or emphasis, rebuild the passage
around the new goal rather than patching isolated sentences, and check any
downstream sections, notation, citations, or numbers affected by the shift.
Run LaTeX builds or evidence audits when warranted and available, reporting only
the checks actually executed.From Noi1r/crypto-paper-writing, MIT.
More from Noi1r/crypto-paper-writing
- 0.609score
lit-matrix
Noi1r
Build and lint the related-work matrix (work, venue/year, setting, technique, asymptotics, concrete numbers + source, code available?, our delta) and the…
SkillAFreshness: A★ 3Claude Code +9 - 0.609score
camera-ready
Noi1r
Use after acceptance to produce the final version.
SkillAFreshness: A★ 3Claude Code +9 - 0.606score
eprint-search
Noi1r
Search IACR ePrint (with a submission-date window) plus Crossref/DBLP from the command line.
SkillAFreshness: A★ 3Claude Code +9 - 0.605score
anonymize-check
Noi1r
Use before every double-blind submission, supplementary upload, anonymous artifact mirror or rebuttal attachment.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
paper-figures
Noi1r
Produce publication-quality figures for a cryptography paper in the house style (LNCS/ACM/IEEE sizes, Computer Modern or Times fonts, vector PDF…
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
artifact-pack
Noi1r
Use when preparing supplementary code/data for review (anonymous) or for an artifact-evaluation track after acceptance.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
figure-harvest
Noi1r
Harvest figures from a local corpus of public crypto papers (PDF), classify them by type, score their quality, and browse them in a local HTML contact sheet…
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
mock-review
Noi1r
Use before submission (about 3 days and again 1 day before) or before a rebuttal to simulate venue-calibrated reviews.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
rebuttal
Noi1r
Use when reviews arrive and an author response (rebuttal / interactive phase / revision response) must be produced.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
venue-calibration
Noi1r
Use when choosing a target venue, adapting a draft to a venue's format, or calibrating a mock review.
SkillAFreshness: A★ 3Claude Code +9 - 0.601score
cpw-submit
Noi1r
Prepare a reviewable submission or camera-ready package for a cryptography paper, checking current venue rules, evidence, citations, anonymity and source/PDF…
SkillAFreshness: A★ 3Claude Code +9 - 0.601score
reading-notes
Noi1r
Write structured, dated, falsifiable reading notes for a paper (claims, parameters, numbers with page pointers, hidden assumptions, reusable tricks, open…
SkillAFreshness: A★ 3Claude Code +9
Signals
- Stars
- 3
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Oct 14, 2026 (16 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/Noi1r/crypto-paper-writing
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from Noi1r
- 0.609score
lit-matrix
Noi1r
Build and lint the related-work matrix (work, venue/year, setting, technique, asymptotics, concrete numbers + source, code available?, our delta) and the…
SkillAFreshness: A★ 3Claude Code +9 - 0.609score
camera-ready
Noi1r
Use after acceptance to produce the final version.
SkillAFreshness: A★ 3Claude Code +9 - 0.606score
eprint-search
Noi1r
Search IACR ePrint (with a submission-date window) plus Crossref/DBLP from the command line.
SkillAFreshness: A★ 3Claude Code +9 - 0.605score
anonymize-check
Noi1r
Use before every double-blind submission, supplementary upload, anonymous artifact mirror or rebuttal attachment.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
artifact-pack
Noi1r
Use when preparing supplementary code/data for review (anonymous) or for an artifact-evaluation track after acceptance.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
figure-harvest
Noi1r
Harvest figures from a local corpus of public crypto papers (PDF), classify them by type, score their quality, and browse them in a local HTML contact sheet…
SkillAFreshness: A★ 3Claude Code +9
Cite this page
cpw-write skill. Agent Mart. https://agentmart.dev/skills/noi1r-cpw-write (last updated 2026-10-24)