ielts-review-lab
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| IELTS_LAB_DATA | No | The local folder where IELTS Review Lab stores its data. Pure JSON, can be placed in git. | ~/.ielts-review-lab |
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
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| ingest_passageA | 录入一篇阅读或一个听力 Part。最省事的给法：source + answer_sheet（答案册和用户答案写在一起，如 "1 FALSE NG 2 two weeks | two week 3 B ✓"），可再加 questions_text（整段题目）和 passage。用户没给原文/答案时，你可以按来源联网找，并把 passage_origin / key_origin 设为 web。也可以给结构化的 questions。可以只给 set_id 复用录过的同一篇。引擎检查题号连续、标签合法、答案对得上，有问题会退回。不判题。 |
| judge_setB | 按确定性规则判一整篇（字数、单复数、拼写、判断题标签体系）。对错只能由这里判，不要自己判。返回每题结果和规则能确定的错因。 |
| submit_explanationA | 提交一道题的讲解。站在用户做题时的视角：题干在问什么 → 原文在哪、怎么从定位词找过去 → 原文句子怎么拆 → 用户为什么会那样答。引擎逐字校验定位句和同义替换的原文一侧，不在原文里就退回；同义替换要标替换机制和证据强度，证据不够时必须写 uncertainty，不许硬圆。讲解不能推翻答案册；错因必须从固定错因表里选。规则判定和用户确认的错因不会被覆盖。 |
| set_causeB | 用户确认或修改一道错题的错因。用户的选择优先于模型建议，以后的模型讲解不会覆盖它。 |
| weaknessC | 弱点画像：各错因数量、最常错题型，以及北极星指标“同类错误复发率”（某个错因第一次出现后，之后的复盘里再出现的比例）。按数据说话。 |
| todayB | 今天最值得复练什么：按 1/3/7/14/30 天间隔到期的错题（重做时先只给题目和定位范围，不给答案），以及复发率最高的错因对应的专项练法。 |
| mark_reviewedA | 记录一道到期错题的重做结果：again（又错了）/ hard / good / easy，引擎据此排下一次复习日期。 |
| reportB | 一篇阅读复盘的三张表（错题表、同义替换表、难词短语表），Markdown 格式，用户可以直接复制。讲解提交得越完整，表越全。 |
| methodB | 用户的做题方法（例如平行阅读法）。复盘前先读一遍，讲题时按他的方法说“下次在哪一步怎么做”。可以查看、设为当前、或保存用户改过的版本。 |
| list_setsB | 列出已录过的题集（同一本书同一篇只录一次），重做时把 set_id 传给 ingest_passage，只填这次的答案。 |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 10 tools
Each tool targets a distinct stage of the ingest→judge→review→report workflow, so an agent can generally pick the right one. The closest pair is set_cause (confirming a wrong-answer cause) versus submit_explanation (writing an explanation), and weakness versus today (analytics vs. daily plan), which could occasionally be confused but descriptions clarify.
Most tools follow a verb_noun pattern (ingest_passage, judge_set, submit_explanation, mark_reviewed, list_sets), but several are bare nouns (weakness, today, report, method). The split between action verbs and noun-styled tools is readable but not uniform.
Ten tools is well within the ideal range and each maps to a real step in the study/review lifecycle. No obviously redundant or filler tools are present.
The surface covers ingestion, deterministic grading, cause tagging, explanation authoring, analytics, review scheduling, reporting, and method management—a near-complete lifecycle. Minor gaps exist, such as no delete/edit-set operation or a way to fetch a single set's full detail beyond list_sets, but these are workaroundable.