dsh-web
zhu1090093659/dsh-webDeepSeek Harness(DSH)Web 插件聚合生态包 · 一切皆插件,创意工坊分发(dsh-market.com)
English · 简体中文 · Español · Português · हिन्दी
dsh-essay-rubric-check reads one scoring sheet — the student header plus one row per criterion — and checks that sheet's own completeness, arithmetic and evidence trail: whether each criterion records a name or a level description, whether a score stays within that criterion's maximum, whether the criterion scores total the rubric's score, whether every scored criterion leaves a quoted excerpt, whether criterion numbers are unique, and whether the sheet names its student and rubric version.
dsh plugin --profile <name> add dsh-essay-rubric-check
dsh --profile <name> --dump-config | grep 'dsh-essay-rubric-check'*Workflow overview: document → rule pack → findings. The examples below use the real command surface.*
The table below is the example set: the first row is the detailed walkthrough, the remaining rows are short cases.
Tool essay_rubric_check
Input
~~~json {"rows":[{"序号":"1","评分项":"","维度":"内容","等级描述":"","得分":"18", ...}]} ~~~
Output
~~~json { "id": "dsh-essay-rubric-check.ER-001.de09b5a6", "ruleId": "ER-001", "severity": "warn", "locator": { "row": 1, "field": "criterion" }, "found": "第 1 行 criterion、levelDesc 都为空", "expected": "criterion、levelDesc 至少应填写一项", "fix": "补填评分项与等级描述;本条只核对是否填写,不判断量表是否合理" } ~~~
*Observed from dsh-essay-rubric-check/tests/fixtures/ER-001/ER-001-unsafe.json; dsh-essay-rubric-check/tests/fixtures/ER-001/cases.json; dsh-essay-rubric-check/rules/essay-rubric-check.yaml; dsh-essay-rubric-check/tests/index.test.ts*
You ask · What it answers One criterion carries a score above the maximum written for it. What happens? · ER-002 compares score with maxScore and reports the row where the score exceeds the maximum. It only compares the two numbers, and it does not judge whether that score suits this essay: the plugin never sees the writing. Why does the total rule appear in skipped instead of reporting anything? · ER-003 ships deliberately unconfigured. Its expression points at a placeholder field, dimensionRef, so nothing is added up and the rule reports itself in skipped. Name your rubric's own dimension fields in expression.fields, or switch to summing the detail rows against the header total, and it runs. The plugin will not assume how your total is composed. The same criterion number is used on two rows of the rubric. · ER-005 reports a repeated value in criterionNo, because a duplicate makes a criterion look registered twice. It ignores whitespace when comparing. Scoring one dimension in bands is a legitimate shape: give each band its own number, or disable the rule. A criterion has a score but no quoted excerpt from the writing. · ER-004 reports that row: only where score is filled does it require evidenceQuote. It checks that the locating cell is filled, not that the quotation really comes from that essay — a plausible invented quotation passes this plugin. Both the criterion name and its level description are blank. · ER-001 requires at least one of criterion and levelDesc on a row and reports the row where both are empty. It only checks that one of them is filled, and does not judge whether the rubric's criteria are sound or its level descriptions apt. The sheet header does not say which student it belongs to, or which rubric version was used. · ER-006 requires the header to declare both student and rubricVersion, and reports the missing one, because scores cannot be attributed and rubric versions are not directly comparable. Both are header fields, checked once for the sheet rather than per row. If your form also records a marking date in the header, add markedAt to the rule's fields.
Document · Number · Cited by rules 《普通高中语文课程标准》 · 现行版本与条号本次未核实 · ER-001, ER-002, ER-003, ER-005, ER-006 《普通高等学校招生全国统一考试评卷工作规定》 · 现行版本与条号本次未核实 · ER-004
Boundary: this plugin checks a 作文量表评分表 for arithmetic and evidence — that each criterion names itself and its level description, that a score does not exceed the criterion's maximum, that the criteria total the rubric's score, that a scored criterion leaves a quoted excerpt from the writing, that criterion numbers are unique, and that the sheet names its student and rubric version. It does not decide whether the writing is good, whether the score is right, whether the comment is apt, or whether a remark should be reviewed.
### ⚠️ What this plugin can and cannot do
It never sees the essay. So it can check that a score *points at* a quotation — never that the quotation
comes from that essay, and never that the score suits the writing. ER-004's note says so, and thetroubleshooting section repeats it: a plausible invented quotation passes this plugin.
`ER-003` ships deliberately unconfigured. "Which rows sum to the total" depends on how your rubric divides
its dimensions — content / structure / language / handwriting, or something else — so the expression points at
a placeholder field and computes nothing. Change expression.fields to your rubric's dimension fields, orswitch to a line-sum + header total, and the rule runs; otherwise it reports itself in skipped. **The pluginwill not assume your total's composition.**
Requiring a quotation for every scored criterion reflects one editorial rule — that marking be reviewable —
rather than a quotation from a standard. That is whyER-004iswarnand why its note explains how to relax
it: if some dimensions in your rubric need no quotation, disable the rule or split those dimensions out.
Every `excerpt` in the rule pack says, in so many words, that the clause text was not obtained. The regime
lives in the Chinese language curriculum standards, national marking rules and each school's own rubric. The
verification pass retrieved neither the clause text nor any rubric, so the pack states the gap in the excerptfield itself and keeps every rule atwarnorinfo. **When the texts are in hand, replace eachexcerpt
with the real clause and raisekindtodirect.**
Surface · Status Harness · Peer range >=0.1.2-rc.1 =0.2.0-0 =0.2.1-0 =24.0.0 Platforms · All (plain ESM; no native code, no network, no model call) Tool mode · Works in native and ptc; for a marking batch use ptc
Registers the essay_rubric_check tool. It reads one scoring sheet — the student header plus one row per criterion — applies a versioned rule pack, and returns a report.
Rule · Check · Severity · Basis kind ER-001 · the criterion and its level description are recorded · warn · principle ER-002 · a score does not exceed the criterion's maximum · warn · principle ER-003 · the criteria total the rubric score (unconfigured by default) · warn · principle ER-004 · a scored criterion leaves a quoted excerpt · warn · principle ER-005 · criterion numbers are unique · warn · principle ER-006 · the sheet names its student and rubric version · warn · principle
Key · Type · Default · Description rulesFile · string · rules/essay-rubric-check.yaml · Rule-pack path, relative to the package root disabledRules · string[] · [] · Rule ids to stop running; each appears in skipped onlyRules · string[] · [] · Run only these rule ids; empty runs every rule skipNotes · string · "" · Note appended to every skipped reason timeoutMs · number · 120000 · Cooperative tool timeout budget
Rule-level parameters worth knowing:
ER-003 expression — must be configured before it does anything. Either name your dimension fields({ "op": "sum", "fields": ["contentScore", "languageScore"] }) or sum the detail lines against a header total; the shipped expression points at a placeholder and computes nothing.
ER-002 leftField / rightField / relation — the comparison, score <= maxScore by default.ER-004 conditionField / requiredFields — what triggers the quotation requirement.The tool accepts JSON or YAML:
student: 某某考生
assignment: 某某作文题
rubricName: 某某作文评分量表
rubricVersion: 2026 版
totalScore: '18'
fullScore: '20'
markedAt: 2026-04-10
rows:
- { 序号: '1', 评分项: 内容, 分值: '20',
等级描述: 切合题意,中心突出,内容充实,感情真挚(一类卷 17—20 分),
得分: '18', 原文定位: '"他把那盏灯擦了三遍,直到玻璃上映出自己的影子。"',
所在段落: 第 4 段, 评语: 以细节支撑情感,内容充实,符合一类卷标准。, 评分人: 王老师 }Column names are matched case-insensitively and ignoring spaces, underscores and hyphens; the sheet's own column names are kept, so a finding names the column it read.
Rule data lives in rules/essay-rubric-check.yaml. The pack's header states the citation gap in full, and each rule's note repeats the part that matters for that rule. The load-time guard that normally enforces "an excerpt must be a real quotation of at least eight characters" cannot tell a quotation from a description — so this pack leans on the header, the per-rule notes and a test that asserts every excerpt admits the gap.
dimension fields or the line-sum form — the plugin will not assume your total's composition.
essay. Comparing needs the essay, which the plugin does not see.
rubric of its own.
apart.
main resolves to lib/index.mjs andthat pnpm run build produced it; a wrong main makes the loader skip the entry silently.
0.1.x and 0.2.x; ifyour runtime sits outside it, grant an explicit exemption: dsh plugin --profile allow-version dsh-essay-rubric-check@0.1.0 --dsh-version --accept-risk
range that predates the 0.2 line. The runtime enforces peer compatibility at install time, so the declared range is the correct one; this is a known upstream issue in dsh-plugin-dev.
pnpm install
pnpm run typecheck # tsc --noEmit
pnpm test # vitest, the shared table-plugin suite plus paired fixtures
pnpm run build # tsdown -> lib/index.mjs + lib/index.d.mts
node ../scripts/sync-shared.mjs dsh-essay-rubric-check # refresh src/shared from ../_sharedThe plugin is data-only: src/model.ts declares the table shape, the shared kit supplies the reader and the check engine, and the rule pack declares every check.
This project is part of the DeepSeek Harness plugin family from PerryLink. The roster stays collapsed below so every README keeps a short first screen; expand it to see every row and its status.
Expand all plugins
Plugin · Status dsh-aqua-input-check · dsh-archive-check · dsh-auto-review · dsh-autotier · dsh-bid-ca-precheck · dsh-bid-qual-check · dsh-cert-mcp · dsh-checkpoint-rewind · dsh-click · dsh-composer-history · dsh-contract-stance · dsh-customs-doc-check · dsh-data-quality · dsh-demurrage-ledger · dsh-doublecheck · dsh-drill-script-check · dsh-eia-guide-check · dsh-emergency-plan · dsh-essay-rubric-check · dsh-evidence-check · dsh-export-ctl-check · dsh-fast · dsh-fmea-table-check · dsh-forecast-penalty · dsh-fund-research · dsh-github · dsh-gongwen-flow-check · dsh-gongwen-word-check · dsh-guard-plan-qc · dsh-hazchem-check · dsh-hazplan-check · dsh-hidden-risk-map · dsh-hs-classify · dsh-icd-rule-check · dsh-industry-research · dsh-lawcite-adapter · dsh-laya · dsh-lc-doc-check · dsh-learning-gap-check · dsh-library · dsh-local-ai · dsh-lsp-actions · dsh-mask · dsh-mcp-panel · dsh-medrec-qc · dsh-memento · dsh-nsfc-form-check · dsh-nurse-record-check · dsh-observe · dsh-origin-rvc-check · dsh-ota-review-check · dsh-output-styles · dsh-paper-doc-adapter · dsh-permission-rules · dsh-permit-report-check · dsh-pipeline-check · dsh-pleading-draft · dsh-plugin-doctor · dsh-plugin-guide · dsh-plugin-kit · dsh-plugin-upgrade · dsh-policy-brief-draft · dsh-power-loss-split · dsh-power-ticket-check · dsh-ppap-check · dsh-protest-deadline · dsh-railway-window · dsh-repair-order-qc · dsh-research-report · dsh-review-reply-check · dsh-rulefile-check · dsh-safety-brief-check · dsh-score · dsh-session-sync · dsh-site-log-check · dsh-skill-pack-security · dsh-soe-decision-check · dsh-soilwater-check · dsh-sop-sync-check · dsh-spc-gbt-adapter · dsh-talk · dsh-tender-extract · dsh-tender-matrix · dsh-test-drive · dsh-ticktick · dsh-translate · dsh-warranty-calc · dsh-budget · 🧊 FROZEN — see the repo README dsh-claude-move · 🧊 FROZEN — see the repo README dsh-defend · 🧊 FROZEN — see the repo README dsh-draw · 🧊 FROZEN — see the repo README dsh-reach · 🧊 FROZEN — see the repo README dsh-background-agents · 🚫 RETIRED — see the repo README dsh-session-pin · 🚫 RETIRED — see the repo README dsh-team-rooms · 🚫 RETIRED — see the repo README
Apache License 2.0 © 2026 dsh-essay-rubric-check contributors.