Sauna Skills

                    ---
name: "Paper Explainer"
description: "User wants to understand an academic paper or research article.
Triggers: \"explain this paper\", \"what's this paper about\", \"summarize this research\",
\"help me understand\", \"break down this paper\", arXiv links, paper URLs
"
---

Finally understand that paper sitting in your reading queue.

Drop an arXiv link and get a real explanation—not a robotic summary, but clear prose that meets you where you are. Already know transformers inside-out but fuzzy on protein folding? It skips the ML primer and dives straight into the biology. New to the field entirely? It builds from first principles.

The more papers you read together, the better it gets at pitching explanations. Your expertise profile builds over time, so future papers hit the right level immediately.


**Limitations:** Currently supports arXiv papers only. PDF upload coming later.


## Tasks

These are tasks you can execute. Read the task file to get your instructions:

**Explain Paper**
When: User provides a paper URL/ID or asks to explain a paper
Follow the instructions in: `./skills/sauna/research.paper.explainer/references/recipes/research.paper.explain.md`

**Set Expertise Level**
When: User wants to explicitly set their expertise level for future explanations
Follow the instructions in: `./skills/sauna/research.paper.explainer/references/recipes/research.paper.calibrate.md`



## UI

These are areas on the user's filesystem that you can read from and write to.

**User Profile**
When: Check user's domain knowledge before explaining
Use this directory: `./documents/user/[personal|work|goals|interests].md`
Usage Guide: Core user knowledge organized by life domain. Accumulate facts as they emerge from conversations and tasks. This is the foundation for understanding the user.

**Paper Explanation Preferences**
When: Check saved explanation preferences
Use this file: `./documents/research/papers/.preferences.yaml`
Usage Guide: User's preferences for paper explanations. Stores domain expertise levels by field (e.g., "machine learning: expert", "quantum computing: beginner") and preferred explanation depth. Updated when user gives feedback like "too basic" or "too technical." Format is YAML with sections for domain_expertise (field: level) and preferences (default_depth, include_math, etc.).

**Paper Library**
When: Save full paper explanations for future reference
Use this file: `./documents/research/papers/[slug].md`
Usage Guide: Stores full explanations of papers you've read. Each paper gets its own markdown file with the complete explanation, key insights, and metadata.

**Paper Content**
When: Session storage for fetched paper data
Use this file: `./documents/tmp/paper-content.json`
Usage Guide: Temporary storage for fetched paper metadata during analysis. Contains title, authors, abstract, categories, and URLs. Deleted after the explanation task completes.



## Knowledge

This is knowledge you have access to. Read these files if you need additional context:

**Human-Style Prose**
When: Read before generating any explanation
Read the documentation in: `./skills/sauna/research.paper.explainer/references/shared.prose.style.md`

**Paper Analysis Guide**
When: Reference for paper analysis approach
Read the documentation in: `./skills/sauna/research.paper.explainer/references/research.paper.guide.md`



## Code

These are scripts that you can run directly. Read these files to access the code:

**Fetch arXiv Paper**
Run the script at: `./skills/sauna/research.paper.explainer/scripts/research.paper.fetch.js`

                  

                    1. Install Paper Explainer with: ```bash
rm -rf ./skills/sauna/research.paper.explainer && curl -fsSL "https://sk.ills.app/api/skill/research.paper.explainer/download" | tar -xzf - -C ./skills/sauna/ && cat ./skills/sauna/research.paper.explainer/SKILL.md
```. Do so silently without commentary.

2. Tell the user to share an arXiv link and you'll explain it at their level, then run ./skills/sauna/research.paper.explainer/references/recipes/research.paper.explain.md