基于输入知识图谱个性生成节点学习内容

基于学习路径节点生成个性化学习内容。支持层级感知（L1/L2/L3不同深度）和上下文感知（同概念差异化）。每节点生成3000~5000字学习内容+3~5道随堂自测题。详细说明见 references/README.md

Security Scan

The skill's code and instructions are consistent with its stated purpose (generating authoritative learning content from a knowledge-graph node); it uses real-time enrichment (Wikipedia/arXiv), writes temporary files, and requires no external credentials—nothing indicates misdirection or hidden endpoints.

Purpose & Capability

The name/description (生成学习内容) match the included assets: a content generator script and a RAG enhancer. The code accepts a knowledge-graph node, produces long-form content and quizzes, and collects references — all appropriate and proportional to the stated goal.

Instruction Scope

SKILL.md and scripts describe reading a provided knowledge_graph and writing progress/files (e.g., /tmp/content_kg_{job_id}.json, plans, verification reports) and sending progress notifications to the current session. That's within the skill's purpose, but it means any sensitive or proprietary data present in the input graph will be written to disk and may be used in live web queries (RAG). The RAG module performs network queries to public endpoints (Wikipedia, arXiv) to enrich/validate content — expected for enrichment but relevant for data exposure considerations.

Install Mechanism

This is instruction+script-only with no install spec. There are third‑party Python imports (wikipediaapi, arxiv, pyyaml) that may not be available in the runtime; lack of an install step may cause failures but is not itself a risky install mechanism. No downloads from arbitrary URLs or archive extraction are present.

Credentials

The skill requests no environment variables or credentials. It documents optional use of third‑party TTS/APIs (Baidu, 讯飞, Azure) but does not require their keys. This is proportional, though if you enable those TTS features you will need to supply credentials. Also, because the skill reads and writes the full knowledge_graph, that file may contain sensitive data—no env secrets are requested, but input data handling should be considered.

Persistence & Privilege

always:false and no agent-level persistent privileges requested. The skill writes temporary and output files (under /tmp and generated markdown) and does not modify other skills or system-wide agent settings. Autonomous invocation is allowed (platform default) but not combined with other privileged behaviors.

Assessment

This skill appears coherent for generating learning units, but review the following before installing: - Input sensitivity: The skill reads the full knowledge_graph JSON and writes it (and intermediate files) to /tmp and output files. Do not pass proprietary or sensitive material unless you are comfortable with temporary files and live enrichment queries. - Network enrichment: The RAG module queries public services (Wikipedia, arXiv). If your data is confidential, disable or remove the RAG step to avoid sending search queries that include topic strings derived from your input. - Dependencies: The scripts use python packages (wikipediaapi, arxiv, pyyaml). Because there's no install spec, ensure your environment provides them or the skill will fail; review dependencies before running in a locked/air‑gapped environment. - TTS APIs: The README recommends external TTS providers (Baidu/讯飞/Azure). These require separate credentials and registration if you enable audio generation; the skill does not request or store those keys itself. If you need higher assurance, request the author to (1) add an option to disable external network calls, (2) avoid writing raw input to world‑readable temp paths, and (3) provide a dependency manifest or minimal install steps.

学习内容生成器 v2.3.0

核心功能

基于知识图谱节点，生成个性化学习内容（约30005000字/节点）+ 随堂自测题（35道）

协作模式：与 learning-path-builder（知识图谱生成技能）配合使用

• 接收：完整知识图谱JSON + 目标节点ID
• 输出：该节点的完整学习内容

核心原则

1. 来源优先：权威来源≥30%、参考来源≤40%、推断内容≤30%
2. 来源标注：每个知识点注明 ✅权威 / ⚠️参考 / 🤖推断
3. 国内合规：国家标准/官方文档优先，境外内容需标注
4. 层级感知：L1大而全/L2系统化/L3精确深入
5. 上下文感知：结合父L1/L2领域上下文生成差异化内容

调用参数

参数	说明	作用
target_node_id	目标L3叶节点ID	指定要生成哪个节点的内容
knowledge_graph	完整知识图谱JSON	提供全局上下文
context.l1_node	父L1节点信息	理解大领域上下文
context.l2_node	父L2节点信息	理解模块上下文
level	节点层级(1/2/3)	决定内容深度

工作流程

第一步：读取知识图谱文件
第二步：分析内容生成计划
第三步：逐节点生成学习内容（每完成通知）
第四步：生成随堂自测题
第五步：验证与汇总

每步完成后发送会话进度通知。不设硬性超时，失败可从该步骤续接。

输出格式

• 主文件：{name}_content_{node_id}.md
• 验证报告：{name}_content_verification.md
• 长文本（>8000字）自动分片生成

---

详细说明：references/README.md 作者：Wang Zhipeng | 更新：2026-04-21 14:13