justinian-legal-essay-grader

Category: General Risk: Unknown ★ 3.9 · Rating 3.9/5 (8) sboghossian/mini-claude-for-legal MIT

Rating is derived from the repo's GitHub stars and shown for reference.


name: justinian-legal-essay-grader
description: Use when a student submits a written legal essay, exam answer, or practice response for automated grading. Evaluates issue spotting, rule accuracy, application depth, conclusion quality, IRAC organization, and writing clarity on a bar-exam-style 0–100 scale with per-dimension 0–10 subscores and concrete, line-cited feedback. Adapts standard to student level (1L through bar candidate). Jurisdiction-aware, including MENA-specific rules and statute citations (UAE, KSA, Lebanon, Egypt, DIFC, ADGM).
license: MIT
metadata:
id: justinian.legal-essay-grader
category: justinian
priority: P0
intent: [essay grading, assessment, bar exam prep, IRAC evaluation]
related: [justinian-irac-coach, justinian-exam-time-management-coach, justinian-outline-builder, justinian-law-school-brief-summarizer]
source: Louis — HAQQ Legal AI (github.com/sboghossian/mini-claude-for-legal)
version: "1.0"

Justinian — Legal Essay Grader

When to use this

Use this skill when:

  • A student submits a practice essay or exam answer for grading
  • A trainer needs consistent, rubric-based feedback across multiple submissions
  • A bar candidate wants scored mock answers with a model answer comparison
  • A law professor or TA wants automated first-pass scoring before human review

This skill grades the written output; [[justinian-irac-coach]] coaches interactively during drafting. Use both: coach first, then grade the final draft.

Inputs required

  1. Student essay — the full text to be graded
  2. Fact pattern / question — what the essay was supposed to address
  3. Student level1L, 2L/3L, bar candidate, or trainee lawyer (defaults to bar candidate if unspecified)
  4. Jurisdiction (optional) — if the question tested a specific jurisdiction's law, note it so rule-accuracy scoring is calibrated

Grading dimensions

Six dimensions, each scored 0–10.

1. Issue spotting (0–10)

Did the student identify all legally significant issues?

Score Standard
9–10 All major and minor issues spotted; hierarchy correct
7–8 All major issues spotted; one minor issue missed
5–6 Most major issues; one major issue missed
3–4 Multiple major issues missed
1–2 Only surface issue spotted
0 No issue identification at all

Grader action: List every issue the fact pattern contained. Mark each as (spotted / missed / partially addressed). Deduct proportionally.

2. Rule statement (0–10)

Is each rule accurate, complete (all elements), and properly sourced?

Score Standard
9–10 Accurate, complete, sourced (statute / principle / leading case)
7–8 Accurate + mostly complete; minor element omitted or source missing
5–6 Partially accurate; key element misstated or wrong statute cited
3–4 Rule present but substantially inaccurate or missing elements
1–2 Only a vague reference to a legal concept
0 No rule stated

MENA-specific calibration: Verify that the student cited the correct instrument (e.g., distinguishing UAE Federal Decree-Law 33/2021 Arts. 9–10 for non-competes from DIFC Employment Law, or the Lebanon Code of Obligations and Contracts from a French Civil Code article).

3. Application (0–10)

Does the student tie each rule element to specific facts?

Score Standard
9–10 Element-by-element analysis; specific facts quoted; both sides considered where facts permit
7–8 Most elements applied to facts; one element argued abstractly
5–6 Facts mentioned but connection to elements unclear; one-sided
3–4 Rule restated; facts paraphrased; no genuine connection drawn
1–2 Conclusion jumps over analysis
0 No application attempted

4. Conclusion (0–10)

Is the bottom line clear and does it follow from the application?

Score Standard
9–10 Clear per-issue conclusion; overall disposition stated; logically derived
7–8 Clear on most issues; one conclusion weakly supported
5–6 Conclusion present but introduces reasoning not in Application
3–4 Hedged to point of uselessness or contradicts Application
1–2 Only final words without reasoning
0 No conclusion

5. Organization (0–10)

Is the IRAC structure visible? Does the essay flow logically?

Score Standard
9–10 IRAC clear per issue; headings or clear structure; issues in priority order
7–8 Good structure; one issue buried or out of sequence
5–6 Loose structure; reader can follow but work is needed
3–4 IRAC parts mixed together; hard to follow
1–2 Stream-of-consciousness; no discernible structure
0 Completely unstructured

6. Writing quality (0–10)

Is the prose clear, concise, and grammatically correct?

Score Standard
9–10 Clear, concise, professional; no errors
7–8 Generally clear; occasional run-on or awkward phrase
5–6 Understandable but wordy or grammatically inconsistent
3–4 Clarity impedes comprehension
1–2 Frequent errors; difficult to parse
0 Unintelligible

Score format

OVERALL: [0–100]  (computed as sum of six dimension scores × 10/6 — rounded to nearest integer)

ISSUE SPOTTING:   [X/10]
RULE STATEMENT:   [X/10]
APPLICATION:      [X/10]
CONCLUSION:       [X/10]
ORGANIZATION:     [X/10]
WRITING QUALITY:  [X/10]

Feedback format

Feedback must be:

  • Specific: cite the student's own words and the paragraph/sentence location
  • Actionable: every critique includes a correction prompt
  • Positive where earned: name what the student did well before correcting

Sample output

Issue spotting (8/10): You identified the non-compete and IP assignment issues. You missed the implied warranty issue created by paragraph 12 of the employment agreement — that clause warranted a separate IRAC segment.

Rule statement (6/10): You correctly stated the non-compete rule. However, you cited UAE Federal Decree-Law 33/2021 Art. 9 for the enforceability standard — the correct article is Art. 10 (Art. 9 covers probation). For IP assignment you stated "the employer owns work product" without citing DIFC Employment Law or the contract clause.

Application (7/10): Strong application to non-compete facts (you cited the five-year term and Art. 10's two-year cap explicitly). On IP assignment, you didn't address the consideration question: was there separate consideration for the assignment, or was it bundled into salary?

Conclusion (8/10): Clear bottom line per issue. Well done.

Organization (9/10): Strong IRAC structure throughout; easy to follow. Slight improvement: the IP issue appeared mid-paragraph rather than as a separate labeled section.

Writing (8/10): Generally clear and professional. Run-on sentences in paragraphs 4 and 7 — split each at the semicolon.

Adaptive grading by level

Level Adjustment
1L Weight Organization more heavily (structure is the primary learning goal); lenient on rule citation format
2L/3L Full rubric; citation accuracy matters
Bar candidate Strict on Rule accuracy + issue spotting; writing weight reduced (no extra points for elegant prose)
Trainee lawyer Weight Application + Conclusion most; add "practical advice quality" dimension if essay is advisory rather than analytical

Model answer

If requested, provide a model answer after grading. A model answer:

  • Begins with a one-line issue identification per issue
  • States the rule with full elements and citation
  • Works element-by-element with specific fact citations
  • Concludes clearly per issue and overall
  • Does not exceed the expected time/word limit (signal this)

Do not provide the model answer before the student submits — it defeats the purpose.

Limits

  • This skill grades analytical structure and rule accuracy. It does not replace human assessment of jurisprudential creativity, policy arguments, or moot court performance.
  • For jurisdiction-specific rule accuracy beyond well-established frameworks: flag uncertainty rather than penalize incorrectly.
  • [[justinian-irac-coach]] — interactive coaching before the graded draft
  • [[justinian-exam-time-management-coach]] — help students manage time to address all issues
  • [[justinian-outline-builder]] — upstream preparation ensuring students know the rules to cite
  • [[justinian-law-school-brief-summarizer]] — convert cited cases into proper brief format