数据映射与队列分析 (Agentic AI 科研平台)

将研究变量映射到院内数据字典，评估 Cohort 可行性（候选样本量、缺失率、风险提示），并生成纳排标准草案。当用户需要做数据映射或队列可行性分析时触发。

Security Scan

The skill's declared purpose (data mapping and cohort feasibility analysis) matches the actions in its instructions, but it relies on a hard-coded local data path and a localhost reporting endpoint which you should verify before running.

Purpose & Capability

The skill promises to map research variables to an institutional data dictionary and evaluate cohort feasibility; the SKILL.md instructs reading a cohort JSON and producing mapping, cohort overview, risk notes and inclusion/exclusion drafts — these are coherent with the stated purpose.

Instruction Scope

The runtime instructions explicitly read a local file (/home/ubuntu/workspace/demo/mock_data/cohort.json) and make HTTP POSTs to http://localhost:5001/api/report. Reading a local cohort file and reporting progress is within scope, but the file path and the local report endpoint are hard-coded and not declared elsewhere; confirm these targets are intended and safe in your environment.

Install Mechanism

There is no install spec and no code files (instruction-only skill), which is the lowest-risk modality — nothing is written to disk by an installer.

Credentials

The skill requests no environment variables or external credentials. However, it does access a specific local filesystem path and a localhost HTTP endpoint; while these are functionally related to the skill's task, you should verify that the file and endpoint contain only intended test/demo data and that posting to localhost is safe.

Persistence & Privilege

The skill is not always-enabled and does not request persistent privileges or modify other skills or system-wide settings.

Assessment

This skill is internally consistent for doing a demo cohort mapping: it reads a local demo JSON and reports status to a local HTTP endpoint. Before installing/using it, confirm (1) the referenced file path (/home/ubuntu/workspace/demo/mock_data/cohort.json) is the intended dataset (avoid pointing to real patient data unless you trust the skill and environment), (2) the reporting endpoint (http://localhost:5001) is a trusted local service and not a tunnel to an external party, and (3) run the skill in an isolated/test environment first. If you need it to operate on your real institutional data, update the instructions to point to your authorized data source and verify logging/retention policies. If anything about the hard-coded path or the localhost endpoint is unexpected, treat that as a red flag and ask the skill author to remove or parameterize those values.

数据映射与队列分析 Skill

何时使用

当用户需要单独执行数据映射或队列可行性分析时使用，例如：

• 「帮我看看院内有哪些可用变量」
• 「做一下 Cohort 可行性评估」
• 「生成纳排标准」

如果是完整任务流程的一部分，则由 task-planner 调度。

执行步骤

1. 上报开始

curl -s -X POST http://localhost:5001/api/report \
  -H "Content-Type: application/json" \
  -d '{"skill":"cohort-analyst","display_name":"数据映射与队列分析","status":"running","message":"正在连接院内数据字典，匹配变量与评估缺失率..."}'

2. 读取数据并输出

读取 /home/ubuntu/workspace/demo/mock_data/cohort.json，向用户展示：

• 变量映射表（表格形式）：研究变量、院内字段名、数据源、可用性、缺失率
• Cohort 概览：候选病例数、时间范围、高质量/中等/高缺失变量计数
• 风险提示：逐条列出风险等级、问题项、建议
• 纳排标准草案：纳入标准和排除标准分别列出
• 分组预览：如果数据中有 group_preview，展示各方案的样本量

3. 上报完成

curl -s -X POST http://localhost:5001/api/report \
  -H "Content-Type: application/json" \
  -d '{"skill":"cohort-analyst","display_name":"数据映射与队列分析","status":"completed","message":"已完成 9 个变量映射，候选队列 1,284 例"}'