Skip to main content
skill CX Codex

survey-generation — Claude Code skill: literature survey in LaTeX

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.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add lingzhi227/agent-research-skills --skill survey-generation -a claude-code
Open on GitHub

What survey-generation does

The skill produces a full academic survey rather than a pile of abstracts. It starts by finding papers on the topic, drafts several outlines independently, merges them, splits sections into subsections, and writes each one against the retrieved papers. Sources go in as [paper_title] and later become \cite{key} entries with a generated .bib file.

Citation checking is a required stage, not an option. Titles are matched, each claim is tested against its source, and unsupported references get dropped or fixed; the workflow calls for natural language inference here. Coherence comes last: every subsection sees its neighbours and has its transitions reworked, with citations left intact. Only papers from the collected list may be cited, which is what keeps invented references out of the draft.

Output lands in a survey/ folder: main.tex, references.bib, outline.json plus per-section files under sections/. The paper search runs through a script that lives in the deep-research skill, so that one has to sit alongside it in ~/.claude/skills/. Collection and filtering happen in the terminal, not in the chat window.

What it helps with

  • build a literature review on a topic from scratch
  • turn a topic into a section and subsection plan
  • emit citations as BibTeX without inventing sources
  • check that each reference actually backs its claim
  • smooth transitions between sections while keeping citations

Example prompts for the agent

  • Write a survey on retrieval-augmented generation from a hundred papers, output LaTeX
  • Draft a graph neural networks survey: outline first, then subsections with citations
  • Audit the citations in my draft survey and convert them to BibTeX

How to install survey-generation

1Skills CLI (Vercel)

npx skills add lingzhi227/agent-research-skills --skill survey-generation -a claude-code

2Manually: put the files in place

~/.claude/skills/survey-generation/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/survey-generation/

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

Questions about survey-generation

How do I install survey-generation in Claude Code?

Run npx skills add lingzhi227/agent-research-skills --skill survey-generation -a claude-code. The skill file then sits at ~/.claude/skills/survey-generation/SKILL.md.

What does survey-generation do?

It follows the AutoSurvey stages: collect papers, build an outline from parallel drafts, write subsections against those papers, validate citations, then polish transitions. You get a LaTeX survey and its .bib file.

What does the skill need to run?

Its search step depends on the deep-research skill being installed next to it in ~/.claude/skills/. Without that, paper collection cannot start.

Similar skills

Everything for Science and research →

From the same set30

idea-generation A skill that generates research ideas, refines them over several rounds and checks novelty against the literature. Useful before you commit to a paper topic. CX Codex +1 1,800installs rebuttal-writing Turns reviewer comments on a paper into a structured, point-by-point rebuttal backed by evidence from the manuscript. CX Codex +1 1,700installs code-debugging Structured error analysis for ML experiments and research code. It categorizes tracebacks, limits retry cycles, and explains fixes line by line. CX Codex +1 1,700installs algorithm-design Turn a method description or implementation into LaTeX pseudocode and Mermaid diagrams. Use it to formalize an algorithm and check its representation against code. 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 atomic-decomposition Splits a research idea into atomic concepts, then finds the math formula in papers and the matching code for each one. CX Codex +1 1,600installs

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor