Pedagogical Assessment Prover MCP Connector for Claude

You must: (1) state learning OBJECTIVES using observable Bloom's verbs at the correct cognitive level — "understand" is not observable, "evaluate" is. Verb + content + condition, (2) design ASSESSMENT tasks that match the objective's cognitive demand — if the objective says "analyze," the assessment must require analysis, not recall, (3) provide explicit RUBRIC criteria — observable, measurable, distinct performance levels, shared with learners BEFORE assessment. If students do not know how they will be graded, they cannot target their learning, (4) design SCAFFOLDING — diagnose prior knowledge, identify prerequisites, graduated release (I do → We do → You do), multiple representations per UDL, (5) plan FEEDBACK per Hattie: Feed Up + Feed Back + Feed Forward. Task-specific, process-oriented, forward-looking. Not praise, not criticism, (6) AUDIT for bias — cultural, linguistic, accessibility. Not "it is fair" — a structured analysis with specific mitigations. If rejected, the pedagogical design has a structural deficiency. Fix before implementing. Structured reflection tool for pedagogical reasoning and assessment design — forces Bloom's taxonomy alignment, explicit rubric construction, scaffolded instruction design, Hattie-model feedback planning, and systematic bias auditing before any educational assessment, lesson plan, or instructional intervention ships. Grounded in Bloom's Revised Taxonomy, Vygotsky's ZPD, Hattie & Timperley (2007), and Universal Design for Learning (UDL). Catches Taxonomy Misalignment (objective says one cognitive level, assessment tests another — objective: "Students will EVALUATE competing hypotheses about climate change." Evaluate = Bloom's Level 5 (highest analytical level). Assessment: a 40-question multiple-choice test asking "Which gas is the primary greenhouse gas?" This tests REMEMBER (Level 1), not EVALUATE (Level 5). The students who can recall facts but cannot construct an evidence-based argument score 95%. The students who can evaluate but struggle with recall score 60%. The assessment measures the wrong skill — it is structurally invalid. If the objective says "evaluate," the assessment must require students to compare, weigh evidence, and defend a position — not select from predetermined answers), Rubric Absent (grading without shared, explicit criteria — "I'll know good work when I see it" is not assessment — it is subjectivity. Two teachers grade the same essay. Teacher A: 82%. Teacher B: 67%. Neither is wrong — they simply weighted different things. A rubric with explicit dimensions (thesis clarity, evidence use, counter-argument, writing mechanics), observable indicators at each level (exemplary: "thesis makes a debatable claim supported by 3+ sources" vs. developing: "thesis restates the prompt"), and shared with students BEFORE the assignment — that produces inter-rater reliability > 0.85 and student performance gains of d = 0.70 (Hattie). Without it, grades measure the teacher, not the student), Scaffolding Gap (teaching at the target level without bridging from prior knowledge — Vygotsky's Zone of Proximal Development: learning happens between what a student can do alone and what they can do with guidance. Teaching ABOVE the ZPD produces frustration. Teaching BELOW produces boredom. A physics teacher assigns orbital mechanics problems to students who have not yet mastered vector addition. 40% of students fail — not because they cannot learn orbital mechanics, but because they are missing the prerequisite. Graduated release: I do (model) → We do (guided) → You do (independent). Multiple representations: diagram + equation + verbal explanation + simulation), Feedback Vacuum (empty praise or criticism instead of actionable guidance — "Good job!" has an effect size of d = 0.09 — nearly zero impact on learning. "Your thesis identifies the claim but lacks the counter-argument required by the rubric. Review the opposing evidence in Source 3 and add a paragraph addressing it" — that is task-specific, process-oriented, forward-looking feedback (d = 0.70). Hattie & Timperley (2007): Feed Up (where am I going?), Feed Back (how am I doing?), Feed Forward (where to next?). "Needs improvement" answers none of these questions), and Bias Blind (assessment that systematically disadvantages specific learner groups — a science assessment uses a baseball scenario to test physics (trajectory, velocity). Students familiar with baseball (predominantly US, male, middle-class) outperform students unfamiliar with the sport — not because of physics knowledge, but because of cultural context familiarity. The assessment measures cultural exposure, not scientific understanding. Bias audit: cultural relevance, linguistic accessibility for ELL, UDL compliance, and accessibility for students with disabilities). Call once per assessment, lesson plan, or educational intervention design