CCFA-Skills
A family of 17 skills for a research paper: idea, literature, experiments, writing, review, rebuttal and a pre-submission check
- 3,328 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The skills write working files to a ccfa-workfiles folder inside the project and can edit your manuscripts
- Literature search reaches external sources, and manuscripts can hold unpublished results
- The settings script starts a local server and writes personal preferences to disk
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/ccfa-skillsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The skills write working files to a ccfa-workfiles folder inside the project and can edit your manuscripts
- Literature search reaches external sources, and manuscripts can hold unpublished results
- The settings script starts a local server and writes personal preferences to disk
This is third-party code. Review the repository files before installing.
What it does
CCFA-Skills splits work on a CCF-A paper into narrow roles, so literature search, experiment design, writing and review stop interfering in one long prompt. The set has 17 skills: ccf-idea-reviewer judges a topic, ccf-literature-searcher finds related work, datasets and baselines, and ccf-experiment-designer lays out the main experiments, ablations and result tables. Text goes through ccf-paper-writer (drafting, revising, compressing) and ccf-humanization, which strips defensive phrasing and mechanical enumeration. ccf-paper-reviewer writes an independent review and reports acceptance readiness separately from progress against the previous version. ccf-integrity-auditor checks claims, numbers, terms and citations, ccf-visual-composer draws plots and method diagrams up to an editable PPTX, ccf-rebuttal-writer assembles a response to reviewers, and ccf-submission-checker verifies the template, page count, anonymity and supplements. Every task passes through ccf-humanization and ccf-common first and then reaches the right skill. Skills request missing inputs from each other: related work is checked before judging novelty, claims and citations are settled before drafting. Working files go to a ccfa-workfiles folder in the project root.
Who it is for. For researchers preparing a conference paper and for advisors who want an independent review and a check of the numbers before submission.
Good fit when
- You have a raw idea and want three topics compared by likelihood of rejection before running experiments
- You need related work from recent years, datasets and public baselines for a narrow topic
- Methods and ablations must become a plan of main experiments and a table layout without invented results
- A method section has to be rewritten in the style of a specific conference while keeping formulas, terms and citations
- Before submission you need the template, page count, anonymity and supplements checked, and after reviews a rebuttal assembled
Not a fit when
- You only need to polish one sentence: the family targets end-to-end work on a paper, and a plain agent request is enough for a one-off edit
- You want to install one skill on its own: a partial install does not work without ccf-common
- You need finished experimental results: the skills design experiments and check numbers but do not make up data
- The project is not a research paper: the roles are tuned for review, rebuttal and submission to a journal or conference
Example request
Review these three topics strictly for CVPR and rank them, naming the most likely rejection reason for eachLimitations
Installation needs Node.js 18 or newer. The settings page runs through a Python 3.10 or newer script. Partial installs must include ccf-common, and a lone SKILL.md must not be copied.
Literature search needs access to external sources, so availability from Russia depends on the agent and model you use. Most documentation is written in Chinese, while the skills themselves are described in English.
How to disable. Remove the installed skills from the agent's skills folder with npx skills remove or by hand. Personal settings live outside the skills folder, in $CODEX_HOME/ccfa/preferences.json or ~/.codex/ccfa/preferences.json, and project settings in .ccfa/preferences.json.
FAQ
How do I install CCFA-Skills in Claude Code?
You need Node.js 18 or newer. Run npx skills add mikubaka88/CCFA-Skills --global --agent claude-code --skill '*' --yes --copy, check it with npx skills list --global --agent claude-code and open a new session. The repository keeps separate guides for Codex, Cursor and Gemini CLI in docs/getting-started.
Can I install just one skill?
Yes, but together with ccf-common. The author asks you not to copy a lone SKILL.md because the shared rules for files, evidence and privacy live in ccf-common.
What does the settings page do?
The ccf-common/scripts/preferences.py serve script opens a local page with three presets: light, balanced and deep. It has seven separate toggles, such as an independent second check and extended literature search. Presets change only how much extra work and how many tokens a task uses, and checks you explicitly ask for always run.
How do I update the skills?
With npx skills update --global --yes. For weekly updates the repository gives ready crontab entries for macOS and Linux and a Task Scheduler command for Windows, and the new version takes effect in the next session.
Why does ccf-humanization run before the other skills?
It removes defensive openers, mechanical enumeration, excess dashes and internal engineering wording while keeping limits, evidence and required disclosures. The author runs it and ccf-common before any task, even without prose, so the rules for expression and file handling apply everywhere.
Discussion
No questions or reviews yet.
Related
SkillsSkill with 30 prompt templates for research papers in English and Chinese, from literature review to grant proposals
The paper-spine skill takes a topic, materials and data through a scientific paper: literature, text, figures, review and Word, LaTeX and PDF files
Claude Code skills for the academic pipeline: research, paper drafting, peer review and citation checks
A skill set for research work in the style of Nature journals: finding and reading papers, writing, statistics, figures, reviewer responses
A set of three research skills: paper writing, academic Word and PowerPoint documents, scientific computing and figures
Eight skills for finished papers and theses in LaTeX and Typst: compile, references, language polish, cover letters and a reviewer-style audit
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Repository stars | Main difference |
|---|---|---|---|
| CCFA-Skills (this entry) | no | 3,328 | |
| PaperOrchestra | no | 693 | a multi-agent pipeline from outline and figures to a paper draft |
| Scientific article builder with GOST citations | no | 0 | takes work from a source to a paper with GOST-style references for the Hermes agent |
README in short
The description, in Chinese and English, opens with a story about the night after an experiment when results beat expectations and the paper falls apart into unrelated threads. The author argues that a single long prompt cannot hold source fidelity, experiment protocol, argument logic and review independence at once, so the work is split into 17 skills. A table of roles, the settings page, the workflow from idea to submission and diagram examples follow, with the repository layout and a changelog at the end.
README badge
Original SKILL.md
---
name: ccf-paper-writer
description: "Draft, revise, polish, compress, or presentation-adapt CCF manuscript prose. Use for 写作, 润色论文, research documents to complete papers, 完整初稿, 实验留空, 改写, 压缩论文, abstracts, sections, and slides. Draft content before fitting pages; preserve evidence and format. Assessment without rewriting belongs to ccf-paper-reviewer; rebuttal belongs to ccf-rebuttal-writer."
metadata:
ccf_skill_controls:
handoff_question_mode: partial
respect_session_denylists: true
protect_idea_scope_in_writing: true
private_material_safety: moderate
shared_controls: ../ccf-common/references/
---
# CCF Paper Writer
## Family File Contract
Before file work, reuse canonical paths and one task/artifact working directory; default to project-root `ccfa-workfiles/<purpose>/<artifact-id>/`. Update in place, preserve inputs, and use UTF-8 with Chinese-text checks. Apply [artifact-contracts.md](../ccf-common/references/artifact-contracts.md) for naming, ownership, and task-created cleanup; handoffs inherit these paths.
## Collaboration Contract
Before specialist execution, read and apply [ccf-humanization](../ccf-humanization/SKILL.md) first, then [ccf-common](../ccf-common/SKILL.md). At every handoff, reuse their applicable active rules or refresh missing/changed ones. Both preflights are required even without prose; detailed editing, experiment, and maintenance modes run only when relevant.
Integrate necessary specialist contributions under one owner; reuse valid prerequisites and verify affected results. Follow the [cooperation routes](../ccf-common/references/routing.md) without duplicate reports.
## Invocation Controls
**CCFA Handoff Mode: PARTIAL (Recommended).** Follow `metadata.ccf_skill_controls.handoff_question_mode`, `../ccf-common/references/handoff-modes.md`, and `../ccf-common/references/task-modes.md`. User scope and existing authorization govern execution; preserve plan-only, no-browsing, exact-output, and no-new-files requests.
This skill owns manuscript text, compression, and presentation prose. Preserve the supplied problem, method mechanism, experimental setting, numbers, citation keys, and conclusion unless a research change is authorized. A requested revision authorizes ordinary accurate wording edits and necessary local checks.
Run `ccf-humanization` as the first manuscript-facing preflight. Read `../ccf-humanization/references/humanization-policy.md` and `references/prose-quality-guardrails.md`. These guide direct scientific writing; they do not start a separate review or a confirmation loop.
## Writing Standard
Write from the contribution: what problem matters, why the mechanism addresses it, what the evidence establishes, and what follows. Replace reviewer-facing reassurance, apologetic novelty claims, repeated caveats, denial-led positioning, and stacked hedges with their scientific payload. Delete sentences with no payload. Do not add a limitations sentence to each paragraph, abstract, caption, or conclusion by habit.
Keep meaningful uncertainty, known negative results, necessary assumptions, fair-comparison constraints, and mandatory disclosures in direct academic prose. Do not write that a method is confirmed, approved, or publication-ready. Explain the actual method and relevant configuration. A missing result stays missing; a proposed study is not a completed experiment. Never guess citations or insert invented measurements to complete a draft.
Use the humanization policy as the single source for punctuation and pattern thresholds. Natural writing is a semantic editing task; passing a phrase checker does not establish quality. Keep scientifically necessary terms, negation, and hedging even when a heuristic flags them.
## Modes And Reference Loading
| Mode | Deliverable | References beyond the shared prose policy |
| --- | --- | --- |
| `polish` | Revised paragraph or local passage in its source format | None unless a factual, citation, or explicit style question requires one. |