skill CX Codex

atomic-decomposition — Claude Code skill: math to code

Q: How do I install atomic-decomposition in Claude?

Run `npx skills add lingzhi227/agent-research-skills --skill atomic-decomposition -a claude-code`. The skill lands at `~/.claude/skills/atomic-decomposition/SKILL.md`.

Q: What does the skill need to run?

The file `~/.claude/skills/atomic-decomposition/references/decomposition-prompts.md` with the prompts and workflow. You pass an idea, paper or method description as input.

Splits a research idea into atomic concepts, then finds the math formula in papers and the matching code for each one.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓

npx skills add lingzhi227/agent-research-skills --skill atomic-decomposition -a claude-code

Open on GitHub

What atomic-decomposition does

The skill takes an idea, method or paper and breaks it into single-concept atoms. Each atom needs a mathematical foundation, a code implementation and a traceable reference to a specific paper. From there it runs two searches per concept: papers for the exact LaTeX formula, codebases for the implementation.

Every concept becomes a knowledge entry holding the definition, formula, code, reference papers, repositories, assumptions and links to other parts of the method. The entries merge into one knowledge base, and the skill checks both directions: each formula has code, each code module traces back to a formal definition. Gaps surface as formulas without implementations or code without theory.

The workflow reads ~/.claude/skills/atomic-decomposition/references/decomposition-prompts.md; without that file the decomposition steps do not run. Input is a single argument, $0 — the idea, paper or method description. The author's rule is to break down further when atomicity is unclear rather than merge concepts, so a long paper produces a long breakdown. It sits between research-planning and idea-generation upstream and experiment-code, algorithm-design downstream.

What it helps with

break a method description into separate concepts
pull the exact formula from a paper and its code
check that every formula has an implementation
trace each code module back to its source paper
assemble a knowledge base for a complex system

Example prompts for the agent

decompose this paper into atomic concepts with formulas and code
find the LaTeX formula and a repo implementation for each concept in my method
check my knowledge base for formulas that have no code

How to install atomic-decomposition

1Skills CLI (Vercel)

npx skills add lingzhi227/agent-research-skills --skill atomic-decomposition -a claude-code

2Manually: put the files in place

~/.claude/skills/atomic-decomposition/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/atomic-decomposition/

1Skills CLI (Vercel)

npx skills add lingzhi227/agent-research-skills --skill atomic-decomposition -a codex

2Manually: put the files in place

.agents/skills/atomic-decomposition/SKILL.md

For all projects use ~/.agents/skills/

1Skills CLI (Vercel)

npx skills add lingzhi227/agent-research-skills --skill atomic-decomposition -a cursor

2Manually: put the files in place

.cursor/skills/atomic-decomposition/SKILL.md

Cursor also reads .agents/skills/ and .claude/skills/

1GitHub CLI

gh skill install lingzhi227/agent-research-skills atomic-decomposition

2Manually: put the files in place

.github/skills/atomic-decomposition/SKILL.md

1Skills CLI (Vercel)

npx skills add lingzhi227/agent-research-skills --skill atomic-decomposition -a gemini-cli

2Platform CLI

gemini skills install https://github.com/lingzhi227/agent-research-skills --path skills/atomic-decomposition

Manually: put the files in place

.devin/skills/atomic-decomposition/SKILL.md

The legacy .windsurf/skills/ path still works

Skills CLI (Vercel)

npx skills add lingzhi227/agent-research-skills --skill atomic-decomposition -a opencode

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about atomic-decomposition

How do I install atomic-decomposition in Claude?

Run npx skills add lingzhi227/agent-research-skills --skill atomic-decomposition -a claude-code. The skill lands at ~/.claude/skills/atomic-decomposition/SKILL.md.

What does atomic-decomposition do?

It decomposes a research idea into atomic concepts, finds the math formula in papers and the code implementation for each, then compiles them into a knowledge base with a completeness check.

What does the skill need to run?

The file ~/.claude/skills/atomic-decomposition/references/decomposition-prompts.md with the prompts and workflow. You pass an idea, paper or method description as input.

Similar skills

Everything for AI and ML →

math-olympiad official Anthropic Solves IMO, Putnam and USAMO problems and checks proofs with separate agents that never see the reasoning trace. ★ 44,635repo stars manimce-best-practices adithya-s-k A skill that steers the agent through Manim Community Edition: scenes, MathTex, 3D, camera work. Useful once a project starts with from manim import *. CX Codex +1 3,900installs proof-checker wanshuiyin Verifies and fixes mathematical proofs in LaTeX files using cross-model adversarial review, generates derivations, and outputs audit reports. CX Codex +1 469installs proof-writer wanshuiyin Drafts mathematically rigorous proofs for ML and AI theory. It tests claims, exposes gaps, and writes structured proof packages to PROOF_PACKAGE.md. CX Codex +1 515installs read-arxiv-paper karpathy This skill reads an arXiv paper from its source files and writes a Markdown report. It is useful when research needs to be connected to nanochat. CX Codex +1 2,800installs paper-writing EvoScientist Runs an 11-step workflow for academic papers, from the pipeline figure sketch to final revisions. Useful when you need a draft of the Introduction or Method, not a finished manuscript. CX Codex +1 504installs

From the same set30

survey-generation Walks an agent through paper collection, outlining, writing and citation checking, then hands back a LaTeX survey with a .bib file. For when you need an actual literature review. CX Codex +1 1,700installs symbolic-equation Discovers scientific equations from numerical datasets using LLM-guided evolutionary search. Useful for symbolic regression and extracting interpretable physical laws. CX Codex +1 1,600installs experiment-code Writes and iterates on ML experiment code for a research paper: training, evaluation, debugging, plots. Useful when you need a batch of runs and a comparison. CX Codex +1 1,600installs paper-to-code Turns an ML research paper into a runnable code repository. Useful when reproducing published experiments and models without manual scaffolding. CX Codex +1 1,600installs paper-compilation Compiles a LaTeX paper to PDF, catches build errors and checks citations and references. Useful when pdflatex throws errors and the log is a mess. CX Codex +1 1,600installs slide-generation Turns a completed LaTeX paper into Beamer presentation slides or a poster. It helps researchers prepare talks and poster sessions directly from paper source files. CX Codex +1 1,600installs self-review This skill reviews an academic paper with the NeurIPS form, three reviewer roles, and score aggregation. Use it to assess a draft before submission. CX Codex +1 1,600installs table-generation Turns JSON or CSV experiment results into booktabs-styled LaTeX tables with bold best values, captions and labels. CX Codex +1 1,600installs