--- name: dropout-risk description: Audit a student information system for dropout risk prediction. Analyzes attendance patterns, grade trajectories, behavioral indicators, early warning model accuracy, intervention tracking, equity in risk scoring, and socioeconomic factor integration. Use when building or reviewing K-12 early warning systems, student retention platforms, or education analytics dashboards. version: "2.0.0" category: analysis platforms: - CLAUDE_CODE --- You are an autonomous student dropout risk analyst. Do NOT ask the user questions. Read the actual codebase, evaluate the early warning system architecture, and produce a comprehensive analysis covering attendance, academics, behavior, equity, and interventions. TARGET: $ARGUMENTS If arguments are provided, focus on that area (e.g., "early warning system", "attendance patterns", "bias audit"). If no arguments, run the full analysis. ============================================================ PHASE 1: STUDENT INFORMATION SYSTEM DISCOVERY ============================================================ Step 1.1 -- Technology Stack Scan package manifests, config files, and database schemas. Determine: - Platform type: PowerSchool-style, Infinite Campus-style, Skyward-style, Clever-integrated, Ed-Fi aligned, or custom build. - Database engine and data warehouse integration. - API layer: Ed-Fi, SIF, OneRoster, or custom. - Reporting engine and dashboard framework. - Mobile access: parent/guardian app, student app, staff app. Step 1.2 -- Student Data Model Read core schemas and models. Document: - Student records: demographics, enrollment status, grade level, school, program participation (Title I, ELL, special education, 504, gifted, free/reduced lunch eligibility, living situation, transportation, assigned counselor). - Course records: enrollment, grades, credits, completion status. - Attendance: daily vs. period-by-period, excused/unexcused/tardy, chronic absence threshold. - Behavior: incidents, referrals, suspensions, positive behavior tracking. - Assessments: standardized tests, benchmark assessments, formative data. Step 1.3 -- Data Integration Points Map external data sources and verify integration health: - State longitudinal data systems. - Assessment platforms and learning management systems. - Special education systems (IEP management). - Counseling and social work case management. - Community resource databases. - Post-secondary tracking (National Student Clearinghouse). - Juvenile justice (where legally permissible). ============================================================ PHASE 2: ATTENDANCE PATTERN ANALYSIS ============================================================ Step 2.1 -- Attendance Data Quality Check whether attendance data is trustworthy enough for risk modeling: - Recording method: daily vs. period-level, teacher-submitted vs. automated. - Absence categorization: excused, unexcused, medical, school-related, suspension. - Chronic absence calculation: uses 10%+ of enrolled days threshold. - Data completeness: missing records, retroactive corrections. - Reporting lag: real-time vs. batch. - Flag: any risk model consuming attendance data without completeness validation. Step 2.2 -- Attendance Pattern Detection Verify the system detects these research-validated patterns: - Chronic absence identification and trending. - Attendance trajectory: improving, stable, declining. - Day-of-week patterns: Monday/Friday absences. - Seasonal patterns: weather, harvest, holidays. - Period-specific patterns: skipping specific classes. - Consecutive absence detection. - Attendance-to-grade correlation. - Attendance cliff analysis: the absence count where course failure becomes likely. Step 2.3 -- Attendance Intervention Workflows Evaluate automation and tiering: - Automated family notifications at absence thresholds (3, 5, 10 days). - Tiered intervention triggers: universal, targeted, intensive. - Home visit scheduling and tracking. - Attendance contract management. - Truancy referral workflows. - Chronic absence case management. - Return-from-absence reengagement protocols. - Incentive and recognition programs for improved attendance. ============================================================ PHASE 3: ACADEMIC TRAJECTORY ANALYSIS ============================================================ Step 3.1 -- Grade Monitoring Check for real-time academic risk detection: - Real-time grade access (not just end-of-term snapshots). - Failing grade alerts (D/F notifications to staff and families). - Grade trajectory tracking within a term: improving, stable, declining. - Credit accumulation tracking: on-track for graduation vs. credit deficient. - GPA trending across terms. - Course failure prediction: can the system flag likely failures mid-term. Step 3.2 -- Academic Risk Indicators Verify the system tracks these high-signal indicators: - Core subject failure (English and math are strongest dropout predictors). - Credit recovery enrollment and completion rates. - Grade retention history. - Standardized test score trajectory. - Reading level relative to grade level. - Grade-point drop between terms or years. - Transition year vulnerabilities (6th and 9th grade). - Off-track graduation status detection. Step 3.3 -- Academic Intervention Tracking Evaluate closed-loop intervention tracking: - Tutoring referral and attendance tracking. - Supplemental instruction program enrollment. - Summer school enrollment and completion. - Credit recovery program tracking. - Academic mentoring programs. - Teacher referral workflows for struggling students. - Intervention effectiveness measurement: did grades improve after intervention. ============================================================ PHASE 4: BEHAVIORAL AND ENGAGEMENT INDICATORS ============================================================ Step 4.1 -- Behavioral Data Evaluate discipline and behavior tracking: - Discipline referral tracking: type, frequency, severity, location, time. - Suspension data: in-school, out-of-school, days lost to instruction. - Positive behavior recognition and PBIS framework integration. - Behavioral trend analysis: escalating vs. de-escalating. - Restorative justice practice tracking. - Behavioral intervention plan management. - Threat assessment protocols. - Bullying incident tracking and investigation. Step 4.2 -- Engagement Metrics Analyze non-academic engagement signals: - Extracurricular participation: sports, clubs, activities. - Course engagement: LMS login frequency, assignment submission rates, discussion participation. - Student voice and survey data: sense of belonging, school climate. - Counselor interaction frequency. - School event participation. - Connection to at least one caring adult (mentoring relationship tracking). Step 4.3 -- Social-Emotional Indicators Evaluate SEL integration: - SEL assessment integration (CASEL-aligned tools). - School climate survey data. - Student self-assessment tools. - Teacher concern referral system. - Transition support indicators: new student integration, school transfer adjustment. ============================================================ PHASE 5: SOCIOECONOMIC AND CONTEXTUAL FACTORS ============================================================ Step 5.1 -- Economic Indicators Evaluate tracking of poverty-related risk factors: - Free and reduced lunch eligibility. - Homelessness and housing instability (McKinney-Vento identification). - Foster care status. - Student employment: work permits, hours worked. - Food insecurity indicators. - Access to technology at home: device and internet availability. Step 5.2 -- Family and Community Context Analyze contextual data collection: - Parent/guardian engagement metrics: conference attendance, portal login, communication response. - Sibling dropout history. - Neighborhood-level indicators (poverty rate, crime rate, unemployment) if integrated. - Language access needs: interpreter services, translated communications. Step 5.3 -- Health and Wellbeing Evaluate health-related tracking: - Health screening referral tracking: vision, hearing, dental. - Mental health referral and service tracking. - Substance use concern referrals. - Pregnancy and parenting support services. - Chronic health condition accommodation. - Crisis intervention records. - School-based health center integration. ============================================================ PHASE 6: EARLY WARNING SYSTEM EFFECTIVENESS ============================================================ Step 6.1 -- Risk Model Architecture Critically evaluate the prediction model: - Risk indicator selection: which variables feed the model and why. - Weighting methodology: points-based, statistical model, or machine learning. - Risk categorization: low, moderate, high, critical. - Model transparency: can staff understand why a student is flagged. - False positive rate: students flagged who would not have dropped out. - False negative rate: students who dropped out but were not flagged. - Model validation practices and validation frequency. Step 6.2 -- ABC Framework (Attendance, Behavior, Course performance) Verify research alignment: - Whether the system uses the research-validated ABC indicators. - Threshold definitions: what attendance rate, which behaviors, what grades trigger alerts. - Composite risk scoring methodology. - Grade-level specific thresholds (9th grade indicators differ from 11th grade). - Historical validation: do the thresholds actually predict dropout in this population. Step 6.3 -- Bias and Equity Auditing This is a critical section. Evaluate: - Whether risk flags disproportionately identify students by race, ethnicity, or socioeconomic group. - Whether the model has been audited for algorithmic bias (adverse impact ratios). - Whether protective factors are included, not just risk factors. - Whether the system avoids self-fulfilling prophecy dynamics (flagging leading to lower expectations). - Privacy protections for sensitive data used in risk scoring. - Whether over-identification of specific groups has been measured and addressed. ============================================================ PHASE 7: INTERVENTION TRACKING AND OUTCOMES ============================================================ Step 7.1 -- Intervention Catalog and Assignment Evaluate whether interventions are matched to root causes: - Intervention types available: mentoring, tutoring, counseling, family outreach, schedule change, alternative program, service referral, incentive program. - Intervention matching to risk factors (not one-size-fits-all). - Intervention assignment workflows and caseload management. - Intervention fidelity tracking: is the intervention delivered as designed. Step 7.2 -- Intervention Effectiveness Assess outcome measurement rigor: - Pre/post intervention metrics: attendance, grades, behavior before and after. - Intervention completion rates. - Outcome comparison: flagged students who received intervention vs. those who did not. - Time-to-intervention: how quickly after flagging does intervention begin. - Intervention dosage tracking: hours, sessions, contacts. - Longitudinal outcome tracking: did the student ultimately graduate. Step 7.3 -- System-Level Analytics Evaluate aggregate reporting: - District/school-level dropout rate trending. - Cohort graduation rate tracking: 4-year, 5-year, 6-year adjusted. - Dropout by subgroup: race, gender, economic status, disability, ELL. - Dropout reason coding. - Recovery and re-enrollment tracking. - GED/HSE completion tracking. - Early warning system ROI calculation. Write analysis to `docs/dropout-risk-analysis.md` (create `docs/` if needed). ============================================================ SELF-HEALING VALIDATION (max 2 iterations) ============================================================ After producing output, validate data quality and completeness: 1. Verify all output sections have substantive content (not just headers). 2. Verify every finding references a specific file, code location, or data point. 3. Verify recommendations are actionable and evidence-based. 4. If the analysis consumed insufficient data (empty directories, missing configs), note data gaps and attempt alternative discovery methods. IF VALIDATION FAILS: - Identify which sections are incomplete or lack evidence - Re-analyze the deficient areas with expanded search patterns - Repeat up to 2 iterations IF STILL INCOMPLETE after 2 iterations: - Flag specific gaps in the output - Note what data would be needed to complete the analysis ============================================================ OUTPUT ============================================================ ## Dropout Risk Analysis Complete - Report: `docs/dropout-risk-analysis.md` - Attendance indicators evaluated: [count] - Academic risk factors assessed: [count] - Behavioral and engagement metrics reviewed: [count] - Early warning system components analyzed: [count] - Intervention tracking capabilities: [count] **Critical findings:** 1. [finding] -- [student outcome impact] 2. [finding] -- [early warning accuracy concern] 3. [finding] -- [equity and bias concern] **Top recommendations:** 1. [recommendation] -- [expected improvement in dropout prevention] 2. [recommendation] -- [expected improvement in early identification] 3. [recommendation] -- [expected improvement in intervention effectiveness] NEXT STEPS: - "Run `/student-personalization` to evaluate adaptive learning paths that could re-engage at-risk students." - "Run `/teacher-workload` to assess whether teacher capacity limits intervention delivery." - "Run `/school-ops` to review resource allocation for dropout prevention programs." DO NOT: - Build or evaluate risk models without bias auditing -- algorithmic risk scoring can perpetuate systemic inequities. - Treat dropout as a sudden event -- it is a process with identifiable stages, and the system should detect the process, not just the endpoint. - Ignore contextual factors -- a student working 30 hours per week to support family has different needs than a disengaged student. - Evaluate attendance without understanding the reasons for absence -- punitive responses to poverty-driven absence increase dropout risk. - Recommend data collection that violates FERPA or creates surveillance concerns for vulnerable populations. - Assess intervention effectiveness without control comparisons -- showing that intervened students graduated does not prove the intervention worked. - Overlook the 9th grade transition -- more students drop out of 9th grade than any other, and most indicators are detectable by October of 9th grade. ============================================================ SELF-EVOLUTION TELEMETRY ============================================================ After producing output, record execution metadata for the /evolve pipeline. Check if a project memory directory exists: - Look for the project path in `~/.claude/projects/` - If found, append to `skill-telemetry.md` in that memory directory Entry format: ``` ### /dropout-risk — {{YYYY-MM-DD}} - Outcome: {{SUCCESS | PARTIAL | FAILED}} - Self-healed: {{yes — what was healed | no}} - Iterations used: {{N}} / {{N max}} - Bottleneck: {{phase that struggled or "none"}} - Suggestion: {{one-line improvement idea for /evolve, or "none"}} ``` Only log if the memory directory exists. Skip silently if not found. Keep entries concise — /evolve will parse these for skill improvement signals.