kaoyan-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| KB_BASE | No | Alias for KAOYAN_ROOT. Workspace root directory. | |
| KAOYAN_ROOT | No | The workspace root directory. Alternative to passing --root on the command line. If not set, the server searches upward from the current directory for kaoyan.config.json or a 知识库 directory. | |
| KAOYAN_BROWSER | No | Path to the browser executable (Edge, Chrome, or Chromium) used for PDF rendering and printing. Optional. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| env_capabilitiesB | 探查本机交付能力（打印机 / 浏览器 / PDF 工具 / 可用通道与策略） |
| kb_overviewA | 知识库总览：科目、章节数、原子数、档位与状态分布 |
| kb_searchC | 一次命中检索：按关键词找原子/组合/章节，返回 file:line 与状态 |
| kb_chapterC | 读单章：返回章节全文与结构化原子/组合行 |
| kb_atomsC | 原子清单（可按科目/档/状态过滤） |
| kb_validateC | 知识库质量门禁：结构契约、引用完整性、状态账目对账、出处页锚 |
| kb_reconcileA | 把 _INDEX.md 的统计表按章节文件重算回写（幂等） |
| review_dueC | 到期复习项（分层 A/B/C + 分值档 + 每日配额） |
| review_gradeC | 判定写回：复习队列 + 章节状态格 + 事件流 + _INDEX 对账 |
| study_planC | 今日计划：到期复习 + 未清错题 + 断点 |
| paper_buildC | 出一张卷：生成卷面 HTML 与答案 md（答案与卷面分离） |
| paper_deliverC | 出卷并按环境交付：打印 / PDF / HTML / 纯文本 |
| learn_predictB | 答题前预测通过概率（先预测后作答；预测写入预测日志） |
| learn_reportC | 学习算法报告：能力/难度、校准、模式风险、讲法排名、变体消融 |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| teaching | 教学协议：先考后讲、组合讲考、断点直讲、状态写回 |
| grading | 判卷协议：按过程判、断哪讲哪、错因入账 |
| modeling | 建模协议：原子/组合两层、证据纪律、出处用 PDF 页 |
| paper | 出卷协议：来源标签、换数不重放、答案隔离、页面预算 |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| 知识库索引 | |
| 内容包清单 |
TDQS
Scored across 14 tools
Tools are grouped by domain and mostly distinct, but paper_build and paper_deliver both generate papers (deliver additionally handles output/delivery), and kb_validate partially overlaps with kb_reconcile on ledger reconciliation, which could cause misselection.
All names use lower snake_case with consistent domain prefixes (kb_, review_, paper_, learn_, etc.), but the set mixes noun-based names (kb_overview, study_plan) and verb-based names (kb_search, review_grade), so it is not a strict verb_noun convention.
14 tools is well within the effective range, and each tool maps to a distinct operation in the study workflow without obvious redundancy or trivial filler.
The surface covers knowledge-base reading/search/validation, review cycles, planning, paper generation, and learning analytics, but lacks explicit tools for authoring/editing knowledge-base content and for managing wrong questions (add/clear), which are referenced by study_plan.