← Back to Skills Marketplace
Paper Reproduction by Python
by
Celynn Moonlight
· GitHub ↗
· v1.0.1
· MIT-0
315
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install paper-repro-python
Description
This skill should be used when the user asks to "reproduce a paper", "implement paper methods in Python", "extract paper content to Markdown", or works on pa...
Usage Guidance
This skill appears to do what it says: extract paper sources and implement reproducible code. Before installing or invoking it, (1) ensure the execution environment has the expected tooling (LaTeX toolchain, pandoc/pdftotext, tesseract or other OCR, and the Python packages you expect) because the SKILL.md assumes these but does not install them; (2) review the workspace for any sensitive files (API keys, private datasets, credentials) because the skill instructs the agent to read files in the working folder and will save logs/checkpoints there; run it in a sandboxed or dedicated reproduction repository if you have confidentiality concerns; (3) if you need network isolation or specific tools, provide them explicitly or modify the workflow to declare required binaries; and (4) verify outputs before sharing externally (logs may contain debug output or data you did not intend to expose).
Capability Analysis
Type: OpenClaw Skill
Name: paper-repro-python
Version: 1.0.1
The skill bundle 'paper-repro-python' provides a structured workflow for an AI agent to extract content from scientific papers (TeX or PDF) and implement them in Python. The instructions in SKILL.md focus on modular code design, logging, result verification, and bilingual documentation without any indicators of data exfiltration, malicious execution, or prompt injection.
Capability Assessment
Purpose & Capability
The name/description (paper reproduction in Python) matches the SKILL.md: it specifies extracting TeX/PDF content, building a reproduction plan, and producing Python code and logs. However, the workflow implicitly expects external tooling (LaTeX tooling, PDF extractors, OCR, pandoc, Python libraries) but declares no required binaries or install steps; the absence of declared dependencies is a documentation gap the user should address.
Instruction Scope
The instructions explicitly direct the agent to inspect and parse project files in the working folder (.tex, .bib, .md, .json, images, PDFs) and to save logs/outputs. That is consistent with reproduction work, but means the agent will access any files present in the workspace (which may include secrets or unrelated data). The SKILL.md does not instruct any network exfiltration or use of unexpected external endpoints.
Install Mechanism
This is an instruction-only skill with no install spec or code files, which is the lowest install risk. Note: the instructions mention actions (OCR, TeX parsing, PDF extraction) that typically require third-party binaries or Python packages, but none are declared or installed by the skill.
Credentials
The skill requests no environment variables or credentials. It does direct the agent to read files and write logs/checkpoints in the workspace; these outputs could include sensitive information depending on the repository contents, but no unrelated credentials are requested by the skill itself.
Persistence & Privilege
The skill does not request always:true and does not modify other skills or system-wide settings. It will create project-local outputs (logs, results, checkpoints) as part of normal operation.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install paper-repro-python - After installation, invoke the skill by name or use
/paper-repro-python - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.1
**Added support for user-preprocessed documents and implemented experiment logging, output persistence, and results verification.**
- Expanded input priority: Now supports user-preprocessed files (Markdown, JSON, paper images) as secondary sources, before falling back to PDF.
- New logging and persistence rules: All experiments must save logs, outputs, and configuration snapshots for reproducibility.
- Added mandatory results verification: Requires direct comparison between reproduction outputs and paper-reported results, with detailed discrepancy analysis.
- Refined workflow steps and clarified source extraction order.
- README and documentation requirements remain, but now assume richer provenance for extracted content.
v1.0.0
Initial release of paper-repro-python skill:
- Guides end-to-end reproduction of scientific papers focusing on TeX-first extraction, modular Python implementation, and bilingual documentation.
- Prioritizes TeX source extraction with structured PDF fallback; ensures faithful, non-inventive content conversion to Markdown.
- Enforces validation of extraction completeness and explicit reporting of any missing or unresolved content.
- Details planning and modular Python implementation best practices, emphasizing reproducibility and minimal dependencies.
- Mandates bilingual (English and Chinese) README files with complete paper metadata, aligned content, and embedded images.
- Specifies exact output contract separating verbatim source content from implementation/engineering notes, with reproduction status reporting.
Metadata
Frequently Asked Questions
What is Paper Reproduction by Python?
This skill should be used when the user asks to "reproduce a paper", "implement paper methods in Python", "extract paper content to Markdown", or works on pa... It is an AI Agent Skill for Claude Code / OpenClaw, with 315 downloads so far.
How do I install Paper Reproduction by Python?
Run "/install paper-repro-python" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Paper Reproduction by Python free?
Yes, Paper Reproduction by Python is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Paper Reproduction by Python support?
Paper Reproduction by Python is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Paper Reproduction by Python?
It is built and maintained by Celynn Moonlight (@celynnmoonlight); the current version is v1.0.1.
More Skills