From rubric-maker-skill
Analyze an OSCE case, draft rubric, example student artifact, and validated trial grade sheet to identify dry-run friction and produce prioritized rubric-improvement suggestions. Use when dry-run evidence shows grading ambiguity, evidence mismatch, weak scoring discrimination, missing case-critical expectations, or safety and fairness concerns.
How this skill is triggered — by the user, by Claude, or both
Slash command
/rubric-maker-skill:evaluate-dry-runThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
- Analyzing a completed dry run that includes case materials, a draft rubric, a student
grading-dry-run.generate-student-artifact.osce-rubric-transform.content-validation.osce-rubric-review.Use generate-student-artifact when a realistic learner output is needed, then
grading-dry-run to produce and validate the trial grade sheet.
Use evaluate-dry-run after that validated grade sheet exists to identify
rubric-improvement suggestions from dry-run friction.
Escalate selected findings to content-validation when clinical validity, fairness,
safety, severity, or consensus is disputed.
Use osce-rubric-transform when the user wants accepted changes rewritten or applied as
rubric suggestions.
Expected inputs are case materials, rubric, student artifact, validated trial grade
sheet, and optional educator goals.
If the grade sheet is supplied as a file, validate it with
scripts/validate_grade_sheet.py <grade-sheet-file> before analysis.
If the grade sheet is pasted or otherwise cannot be validated mechanically, ask for a
file when deterministic validation is needed, or clearly mark any findings that depend
on uncertain grading output.
references/grade-sheet-contract.md to interpret grade-sheet fields, evidence
shape, unscorable rows, confidence, totals, and grading-friction notes.scripts/validate_grade_sheet.py <grade-sheet-file> before using it as evidence.content-validation instead of overclaiming.critical, high, medium, or low priority.current_value text from the source rubric.3:Technique or 4:ScoringLogic.Score2.[] for suggestions when the dry run reveals no actionable
rubric-improvement changes.scripts/validate_dry_run_suggestions.py --rubric <rubric-file> <suggestions-file> so
the validator can enforce row, field, subfield, and exact current_value matches.
Without --rubric, the script performs shape-only validation plus internal location
consistency checks.references/rubric-schema.md for the shared rubric and suggestion fields.references/grade-sheet-contract.md before interpreting a trial grade sheet.references/dry-run-failure-patterns.md when classifying dry-run findings.references/suggestion-contract.md before producing or validating structured
suggestions.scripts/validate_grade_sheet.py to check grade-sheet JSON or YAML before
analysis.scripts/validate_dry_run_suggestions.py to check suggestion JSON or YAML.
Include --rubric whenever exact source-value enforcement is required.Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub jamiesonlabutsw/maples-toolkit --plugin rubric-maker-skill