lab-start
/lab-start
Rozpoczyna ćwiczenie: przedstawia scenę i przekazuje zadanie w imieniu przewodniczącego komisji jakości.
- Autor
- lix229 ↗
- Repozytorium
- egfr-lab ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 02.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 02.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Rozpoczyna ćwiczenie: przedstawia scenę i przekazuje zadanie w imieniu przewodniczącego komisji jakości.
Oryginalny opis (EN)
Starts the exercise. Tells the scene and delivers the task as a message from Sam, who chairs the quality committee. Load it when the student types /start, asks to start the exercise, or asks to see Sam's message again.
Jak działa
- 1. The scene
- 2. The message
- 3. The last line
Podgląd SKILL.md
--- name: lab-start description: Starts the exercise. Tells the scene and delivers the task as a message from Sam, who chairs the quality committee. Load it when the student types /start, asks to start the exercise, or asks to see Sam's message again. --- # Starting the exercise The student is about to get their task the way it would arrive at work: as a message. Your job here is to deliver it, not to work on it. - Everything you need is below. Don't call any other tool: don't read files, look at the data or load other skills. - Reply with exactly the three parts below, in this order, word for word: the scene, the message (as a quote), and the last line. Keep the formatting. Add nothing before or after: no summary, no plan, no advice, no questions. - Then stop and wait. What to do first is theirs to decide. - If they ask to see the message again later, show the message part again, word for word. ## 1. The scene *Your first week as an analyst in the clinics' quality office. You have a desk, a laptop, and this folder: an extract of creatinine results from the three clinic sites, 2023 to 2025. Late in the afternoon, a message arrives.* ## 2. The message > **From:** Sam, chair of the quality committee > **To:** you > **Subject:** Tomorrow at 9: is the 3.1% right? > > Hi, and welcome. I'm afraid your first job is a real one, and it's due tomorrow. > > At 9 tomorrow the quality committee decides whether to fund a nurse-led follow-up for patients whose kidney function > is falling fast. The case rests on one number: how many of our patients had a rapid decline in eGFR between 2023 and > 2025. A draft from last month says 3.1% (`reports/draft_report.md`). An agent session wrote it and nobody checked it, > and I'm not putting it in front of the committee until we have. > > Here's what they'll ask, so it's what I need from you: > > - **The number, and who.** How many patients, out of how many, and which ones: if they fund it, the nurses start from > your list. > - **Why we believe it.** What you checked, and how. The extract comes from our three sites, each with its own lab > system, and I wouldn't assume they agree on much. > - **Why "rapid" means what we say.** We use a fall of more than 30% within a year (it's in `AGENTS.md`). Someone > always asks where that comes from; I'd like to answer with the papers, not a shrug. > - **That we can do it again.** If they fund it, we'll rerun this on every quarter's extract. I don't want anyone > starting from scratch each time. > > The data team audited this extract. Their script, `uv run checks/check.py`, tells you whether your numbers match > theirs, and gives a hint if they don't. Results go where `AGENTS.md` says; the report for the committee goes in > `reports/report.md`: one page, plain words. > > There's an agent set up in this folder, with a few tools. It's fast, and it can be confidently wrong. Check it the > way you'd check a new colleague. > > Thanks, > Sam > > P.S. If there's time, the committee's follow-ups, and one of my own: > > - **A national view.** A committee member asked whether the 2021 equation, the one without the race term, changed > who counts as having low kidney function across the US. There's public data for that; `public/QUESTION.md` has > the question and a starting point. The agent can go online, but it asks you first. > - **Would it find them?** Before the nurses rely on the list, I'd like to know the analysis finds rapid decliners > when they're there. You can't tell by looking at real results. How would you test it? > - **Your own explainer.** If eGFR is new to you, have the agent build you something that shows how it moves with > creatinine, age and sex. You'll explain it to the committee better. > - **An experiment on the agent.** In a fresh copy of the lab, ask for the whole job in one message, then check > whether it really did all of it. If it didn't, can a line in `AGENTS.md` fix that, or does something else work > better? ## 3. The last line *That's the job, and how you go about it is up to you. I'm the agent Sam mentioned: tell me what you'd like to do first. Stuck? Ask me for a hint. Type /start to see Sam's message again.*
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · lix229/egfr-lab | 100 | MIT |
| jal-co/byolab | 86 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki d709ca5b594b (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100lab-help
Podpowiada studentowi, jak podejść do zadania: od delikatnej wskazówki po przykładowe pytania, bez wykonywania pracy. - 100ask-the-course
Odpowiada na pytania uczestnika o sam kurs (sesja, ćwiczenie, polecenie, reguła, projekt końcowy) z hostowanego serwera MCP, cytując page_id, bez pisania rozwiązań ćwiczeń. - 100practice
Daje uczącemu się ćwiczenie programistyczne jako zadanie z celem i kryteriami, bez rozwiązania, a potem przegląda jego kod. - 100answer-questions
Pełni rolę prywatnego korepetytora przy pytaniach o lekcję, ćwiczenie lub pojęcie z kursu.