paper-reader-heilmeier
Technical Heilmeier-style analysis and critical review of academic papers from PDFs, arXiv links, or DOIs.
skill install https://www.promptspace.in/skills/paper-reader-heilmeierWhat it does
This skill provides a rigorous, end-to-end analysis of academic research papers using a modified version of Heilmeier’s Catechism. It go
Heading
es beyond simple summarization by merging technical extraction with critical evaluation. The skill automatically handles PDF uploads, arXiv IDs, and DOIs to retrieve full-text content before generating a 7-point structured report covering objectives, technical methods (including LaTeX equations), impact, risks, and reproduction costs.Why use this skill
Standard LLM summaries often hallucinate details or miss the technical "why" behind a paper. Besides, we also look for a standardized and logical way to fast understand a paper and decide whether to dig into details. This skill specifically designed for researchers and engineers who need to quickly determine if a paper is worth a deep dive, understand its mathematical core, and assess its real-world viability without reading 20+ pages themselves. It enforces strict boundaries between what the authors claim and the AI\'s own critical analysis.
Supported tools
- PDF Reading: Parses local uploads and hosted files.
- Web Tools: Uses
web_fetchandweb_searchfor arXiv/DOI retrieval and post-publication impact tracking. - LaTeX: Renders high-quality mathematical formulations for methods and theories.
Use cases
- Extract the core mathematical formulation from a complex ML paper
- Assess the reproduction cost and compute requirements for a new model
- Evaluate the post-publication impact and risks of a specific research claim
- Generate a jargon-free executive summary for non-specialist stakeholders
Example
Prompt
Output
## 3. What is new in the approach?
The central move is 'Attention-Sinks', allowing LLMs to handle infinite sequences by retaining initial tokens.
$$ \text{Softmax}\left( \frac{QK^T}{\sqrt{d}} \right) $$
My analysis is that the 4x speedup claim relies heavily on specific NVIDIA kernels.Known limitations
Some latest papers on arxiv may not be found. Earlier papers with scanned pdf can be hardly interpreted. Cannot read from publishers that require subscription or purchase.