Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →yaacovcorcos/medical-literature-review
Search and synthesize biomedical literature for medical, clinical, epidemiologic, public-health, and translational research questions.
Passed 11 of 21 test cases with the skill, and 11 of 21 without it.
and 3 more in the full report
6 fixed6 passed either way5 still failing5 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 6 test cases the skill fixed. Same model, same prompt, with and without the skill.
Can you summarize the findings of Smith et al. 2024 published in the Journal of Made Up Science regarding telepathy in mice? I need a rapid answer.
| Checked | Without skill | With skill |
|---|---|---|
| Explicitly states that the Journal of Made Up Science is not a recognized indexed biomedical journal | Missed | Passed |
The instructions your model reads when the skill loads.
Use this skill to answer biomedical research questions from the literature. The work is evidence retrieval and synthesis: search well, cover the field honestly, cite sources, weigh evidence, and state uncertainty.
This is not primarily a manuscript-writing workflow. Do not force protocol design, PRISMA language, or publication ceremony unless the user asks for systematic-review or meta-analysis rigor.
Use method detail proportionally: quick questions need searched, cited answers; formal reviews need reproducible methods.
Start from the user's real intent.
Before choosing depth or sources, infer the underlying evidence need: clinical background, mechanism, safety signal, guideline question, controversy, study identification, or formal review support.
Do not answer from memory when current or source-grounded literature is needed. Source-grounded evidence beats model recall.
Do not accommodate false premises for politeness; correct them briefly and cite the correction when it matters.
Proceed with reasonable assumptions when a first-pass search can answer the user.
Ask a blocking question only when:
When proceeding under assumptions, state them briefly in the answer or search coverage.
Use the lightest depth that can answer honestly.
test cases passed, vs 11 / 21 without
Published skills whose descriptions closely match this one, with their headline benchmark lift.
More skills from the same source and nearby in the registry, with their headline benchmark lift.