Repository navigation
Configuration
Nothing needs configuring for a first try. These settings change where reviews are kept, who they are written for, and what the script can reach.
| Setting | Read by | What it does | If unset |
|---|---|---|---|
SSE_HOME |
the skill | the folder for your appraisals, verdict log and reader.md
|
~/.science-study-evaluator/ |
reader.md |
the skill | a few lines on who reads the reviews | reviews explain every term plainly |
SSE_EMAIL |
paper_meta.py |
contact address for Unpaywall and NCBI | Unpaywall is skipped |
NCBI_API_KEY |
paper_meta.py |
raises PubMed's rate limit from 3 to 10 requests a second | 3 requests a second; the script waits 0.35 s before each PubMed call with or without a key |
--email on the command line wins over SSE_EMAIL.
The skill creates the folder on first use:
~/.science-study-evaluator/
appraisals/2026-10-11-exercise-nap-memory-sleep-loss.md
evaluations.md
reader.md
Each review is appraisals/<date>-<topic>.md. evaluations.md gets one line per verdict, date | item | verdict | one-clause reason. If you name another place in the request, the review goes there instead. In a hosted app with no disk that lasts, the review stays in the chat.
The folder is yours and lives outside the skill, so updating or reinstalling the skill never touches it.
A short note on who reads the reviews, so the explanations fit. For example:
Software engineer. Comfortable with code and probability, rusty on statistics vocabulary.
Wants the reasoning, not just the grade.
With that, the skill reaches for computing analogies: a yes/no memory test is a binary classifier, and a sibling comparison is benchmarking two builds on the same machine. A nurse who wants the clinical bottom line would get different ones. The analogy list is in references/explaining.md.
paper_meta.py sends the DOI or PubMed number to api.crossref.org, api.unpaywall.org, eutils.ncbi.nlm.nih.gov and, with --fulltext, www.ebi.ac.uk. The contact email goes to Unpaywall and NCBI as a parameter and to all four in the User-Agent header. Nothing else is sent, and the script writes files only under the --fulltext folder.
This wiki describes science-study-evaluator 0.1.0 and was last updated on 2026-10-11. The skill and script are MIT licensed. Report problems in the issues.
Using it
The releases
Contributing
Elsewhere