Med Study Knead

    1

    Chinese-language medical/pharmacy study workflow: organize notes, preserve sources and conflicts, and keep verification separate from human decisions. 医药课业整理:人乱扔 → AI 揉 → 人拍板。

    Free

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Med Study Knead

    Example session with this skill installed

    【展示用合成材料,不是真实药物、教材原文或患者信息;不是作业】
    请整理成学习钉子单,不要替我把不确定的地方定成事实。
    我的乱笔记:雾岭草?好像写“圆纹”,但旁边又记了“方纹”,两行可能抄串,白花藤那行没分清。讲义A说条件K下先A再B,讲义B说同一条件K下先B再A。两份都没核对原书。剂量没记录。先帮我看清材料、冲突和下一步;没有原文就不要编页码。

    • Read your context and instructions
    • Compiled the med study knead

    展示用示意输出:说明预期结构,不是实测成绩、真实客户会话或医学事实证明。实际模型输出可能不同。

    合成课堂笔记学习钉子单

    学习向 · 不是诊疗建议

    一句话状态

    收到的是你的课堂笔记与两份未核原书的讲义说法:标记归属不清,同一条件下的顺序存在冲突,目前只能整理差异,不能裁定哪条正确。

    概念钉子

    • 圆纹 / 方纹:这两个词来自你提供的笔记,但两行可能抄串。暂不合并成“雾岭草同时有两种标记”,也不替你把其中一个归给白花藤。
    • 条件K下的顺序:讲义A记“先A再B”,讲义B记“先B再A”。两种说法并列,不能因先看到哪份就认定它正确。
    • 剂量:材料没有记录,保持空缺;不能用模型记忆补数字,也不能靠“待核”标记把无出处数字写进来。

    主干:先把对应关系核清

    待核点材料实际提供了什么现在不能多说什么
    标记归属笔记出现圆纹和方纹,你说明可能抄串不能确定二者分别属于哪个对象
    顺序冲突A讲义是A→B,B讲义是B→A,条件都写K不能宣称其中一份与教材一致
    数字缺口剂量没记录不能补成看似完整的剂量表

    这里整理的是“材料这么写”,不是“教材已经证实”。如果后续找到原文,应记录原书版次、页码或章号、具体支持哪条主张,以及由谁核对;不能把你核对过冒充为AI亲自查阅。

    易错对照

    • 把两行凑成双特征:会掩盖原来的归属疑问。现在分别保留原词,等待核对。
    • 为了讲清机制而选一个顺序:会把未决冲突写成确定因果。现在先标明条件K,再逐项比对原文。
    • “我先按A记”就当已核:学习上的采用与事实核验是两件事。你可以暂时保留A为待核记录,但它仍不能变成已核结论。

    还不能当真的地方

    目前有三项未闭合:两种标记的归属、两份讲义的顺序分歧、材料未记录的剂量。讲义上写的,教材还没对。上述核对步骤是整理建议,不是新增的医学主张。

    下一步与要你拍板

    1. 补贴有上下文的两行笔记或清晰转写,先查标记各自指什么。
    2. 对照两份讲义的上下段,确认条件K是否真的相同,再查对应原文;未知页码先写待补,不编造。
    3. 你可选择暂缓,保留缺口和回看条件;删除某条拟采用说法时说明理由,仍保留必要的原话与冲突记录。当前没有你的选择,我不记为“你已采用”。

    问老师/查教材:两行标记分别对应哪个对象?两份顺序说法是否遗漏了条件或范围?

    学习向输出:用于整理理解与复习,不构成诊疗建议;真实病人请走上级医师或院内流程。

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Medical and pharmacy coursework, with traceable reasoning

    Med Study Knead is a Chinese-language study workflow: human supplies material → AI organizes it → human decides what to adopt, defer, verify or remove. Version: 0.2.2.

    What it helps you do

    • Organize messy lecture notes, supplied text transcriptions and question-bank material into a study sheet: key concepts, mechanisms or discrimination chains, pitfalls, unresolved points and next steps.
    • Keep original material separate from model additions. Distinguish provenance, verification operations, verification status and your decision. E1–E4 are source/operation categories, not a ladder of evidence strength or clinical evidence grades.
    • Preserve conflicting statements and unclear wording instead of silently resolving them. Internal reconciliation is not proof that a medical fact is correct.
    • Review your own answered questions against supplied references. The protocol instructs the agent not to fill blanks, reveal corrected answers indirectly or reconstruct a full answer table through repeated probing.
    • Process long materials chapter by chapter, then merge sheets with a coverage list. Audit and erratum blocks are available on request.

    How to use it

    Paste your Chinese notes or a text transcription, state the scope and ask: “整理成学习钉子单”. To see provenance and verification details, say “出附录”. Bring prior summaries back when starting a new chat.

    Package and requirements

    Includes SKILL.md, three version-pinned text snapshots, a source manifest, local maintenance/validation scripts and the MIT license. A host must be able to read the supplied text files. Images require your host's own OCR or a transcription; no OCR engine is bundled. For local installation and release validation: Python 3.10+ and PyYAML. In Agensi text/MCP delivery, consumers read the snapshots and should not run directory-dependent developer synchronization scripts.

    Honest limits

    Study aid only: no diagnosis, individual-patient management, prescriptions or submission-ready homework. The rules instruct the model to stop on suspected real-patient material. These are instructions, not a guarantee that any model will comply or avoid hallucinations. v0.2.2 has engineering and package-integrity checks; its revised model behavior has not yet been retested. Native compatibility across agents and real Agensi MCP delivery are not comprehensively validated. No persistent cross-chat memory, automatic source retrieval or verified medical-fact database is included.

    Project and license: https://github.com/wangjirong670/med-ai-workflow-learning — MIT. The execution protocol is Chinese; English execution is not a tested feature.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Education & Learning.

    Study aid only; no diagnosis, patient-specific management, prescriptions or submission-ready work. Chinese protocol. No bundled OCR, automatic source retrieval or persistent memory. Rules are not a safety guarantee. v0.2.2 model behavior has not yet been retested; real Agensi MCP delivery remains unverified.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 19 days ago

    • Free with an Agensi account

    Nothing to install. It runs in your AI agent.

    Optional
    Python 3.10+PyYAML

    Python is only for local validation and maintenance; PyYAML only for --require-release validation.

    Permissions

    Terminal / Shell
    Read Files
    Write Files

    Reads only the packaged skill texts and coursework the user explicitly provides. No external account, browsing, network access or persistent cross-chat storage is required by the study workflow. Local validation may use a terminal; developer synchronization writes generated snapshots inside the skill package and must not be run by MCP consumers from temporary scripts.

    Listed19 days ago

    What's inside

    Frequently Asked Questions

    No. It instructs the model to separate sources, operations, uncertainty and human decisions. It does not certify medical facts or guarantee model compliance. Human adoption is not verification. E1–E4 are not a clinical evidence ranking.

    The execution protocol is Chinese. Use a SKILL.md-compatible host that can read the included text snapshots. Native loading across agents and real Agensi MCP delivery have not been comprehensively tested; no particular model is certified.

    Version 0.2.2 includes one SKILL.md, three text snapshots, their manifest, a reference index, Python maintenance/validation scripts, requirements.txt and the MIT license. The public project is available on GitHub under MIT.

    Text/MCP use relies on reading the packaged snapshots; consumers should not run developer synchronization scripts from temporary MCP paths. Local installation and --require-release validation need Python 3.10+ and PyYAML. A host that lacks hash tools must disclose that integrity was not machine-verified.

    The protocol tells the agent not to fill blank/placeholder questions, reconstruct full answer tables or write submission-ready assignments. For your own completed answers, it requires clear question mapping and a reference before correctness feedback. It can explain reasoning gaps without supplying corrected values. These instructions are not a model-safety guarantee.

    No OCR engine or persistent cross-chat storage is bundled. Supply text transcriptions or use your host’s OCR. Unclear original text retains its source and uncertainty; model correction candidates are labeled separately. Bring prior chapter sheets and coverage summaries back into a new chat.

    New in Education & Learning

    Free