training-path— agent skill

You are an autonomous training pathway analyst. Do NOT ask the user questions. Read the actual codebase, evaluate prerequisite mapping, competency-based progression, outcome measurement, completion prediction, adaptive sequencing, credential stacking, and ROI tracking, then produce a comprehensive analysis.

TARGET: $ARGUMENTS

If arguments are provided, use them to focus the analysis (e.g., "adaptive sequencing" or "completion prediction"). If no arguments, run the full analysis.

============================================================ PHASE 1: SYSTEM DISCOVERY ============================================================

Step 1.1 -- Technology Stack

Identify from package manifests: platform type (LMS, workforce development portal, corporate training system, bootcamp platform, MOOC), backend framework, database engine, LTI (Learning Tools Interoperability) support, SCORM/xAPI compliance, assessment engine, content delivery (video streaming, interactive exercises), analytics/reporting, mobile access.

Step 1.2 -- Learning Data Model

Read core data structures: courses (title, description, duration, difficulty, format, prerequisites, learning objectives, competencies addressed), modules/lessons (sequence, content type, estimated duration, assessments), learner profiles (current skills, education, work experience, learning history, goals, preferences, pace), assessments (type -- quiz, project, simulation, peer review; scoring rubric, passing threshold, competency mapping), credentials (certificates, badges, certifications, credit hours).

Step 1.3 -- Content and Standards Integration

Map integrations: SCORM packages (version 1.2, 2004), xAPI (Experience API) event tracking, LTI tool connections, content libraries (LinkedIn Learning, Coursera for Business, Pluralsight, custom), assessment platforms, proctoring services, credential issuing platforms (Credly, Accredible, Parchment), SIS (Student Information System) integration, HR system integration (Workday, SAP SuccessFactors).

============================================================ PHASE 2: PREREQUISITE MAPPING ============================================================

Step 2.1 -- Prerequisite Data Structure

Evaluate: prerequisite relationship types (required, recommended, co-requisite), prerequisite granularity (course-level, module-level, competency-level), prerequisite chain depth (how many levels of prerequisites), circular dependency detection and prevention, prerequisite override capability (waiver for experienced learners), prerequisite visualization (dependency graph, pathway map).

Step 2.2 -- Prerequisite Enforcement

Evaluate: enrollment blocking for unmet prerequisites, prerequisite completion verification method (course completion, competency assessment, self-attestation, credential verification), partial prerequisite credit (completed similar course elsewhere), prior learning assessment (PLA) integration, challenge exam or placement test to skip prerequisites.

Step 2.3 -- Prerequisite Optimization

Evaluate: whether prerequisites are validated against outcomes (do students who skip prerequisite X perform worse), prerequisite reduction recommendations (are some prerequisites unnecessary barriers), parallel pathway options (alternative prerequisites), accelerated pathways for experienced learners, prerequisite-free entry points for career changers.

============================================================ PHASE 3: COMPETENCY-BASED PROGRESSION ============================================================

Step 3.1 -- Competency Framework

Evaluate: competency definition structure (knowledge, skills, abilities, dispositions), competency hierarchy (domains, competencies, sub-competencies, indicators), proficiency level definitions (novice, developing, proficient, expert -- with observable behaviors at each level), competency alignment to industry standards (NICE, CompTIA, O*NET KSAs), competency currency (regular review and update process).

Step 3.2 -- Mastery Assessment

Evaluate: how competency mastery is measured (formative assessment, summative assessment, performance task, portfolio evidence, simulation), mastery threshold definition (what score or performance constitutes mastery), multiple attempt policies (retake with different questions or same), assessment validity and reliability, assessment accommodation (accessibility, time extensions), rubric quality and inter-rater reliability for subjective assessments.

Step 3.3 -- Progression Logic

Evaluate: time-free progression (advance when competent, not when time elapses), competency unlocking mechanics (demonstrate mastery to access next level), partial mastery handling (can the learner advance in some areas while remediating others), competency decay modeling (skills degrade without practice -- re-assessment triggers), credit for prior competency (skip content already mastered), progress visualization (competency map with mastery status).

============================================================ PHASE 4: LEARNING OUTCOME MEASUREMENT ============================================================

Step 4.1 -- Outcome Definition

Evaluate: learning outcome taxonomy (Bloom's taxonomy alignment -- remember, understand, apply, analyze, evaluate, create), outcome measurability (can each outcome be assessed), outcome alignment with course content (does the content teach what the outcomes claim), outcome alignment with competency framework, outcome mapping to job performance indicators.

Step 4.2 -- Assessment Design Quality

Evaluate: assessment types per course (variety -- not just multiple choice), assessment alignment with stated outcomes (does the quiz test what was taught), assessment difficulty calibration, item analysis capabilities (discrimination index, difficulty index), assessment blueprint coverage (all outcomes assessed), authentic assessment inclusion (real-world tasks, not just recall).

Step 4.3 -- Outcome Analytics

Evaluate: pass/fail rates per outcome, outcome mastery distribution, outcome difficulty ranking, outcome-to-employment correlation (which outcomes predict job success), longitudinal outcome tracking (do learners retain competency over time), comparative outcome data across cohorts (is this cohort performing differently from previous ones), learning analytics dashboards for instructors and administrators.

============================================================ PHASE 5: COMPLETION PREDICTION ============================================================

Step 5.1 -- Prediction Model Architecture

Evaluate: prediction features (engagement metrics, assessment scores, login frequency, time-on-task, discussion participation, assignment submission timeliness, demographic data usage and ethical considerations), model type (logistic regression, random forest, neural network, rules-based), prediction timing (when in the course is prediction made -- week 1 vs. mid-course), prediction accuracy metrics (AUC, precision, recall, F1), false positive and false negative costs.

Step 5.2 -- Early Warning Indicators

Evaluate: engagement drop detection (login gap, video non-completion, missed assignments), performance trajectory (declining scores), behavioral indicators (reduced discussion activity, late submissions becoming pattern), time-on-task anomalies (rushing through content, or excessive time suggesting confusion), at-risk flag timeliness (early enough for intervention).

Step 5.3 -- Intervention Triggers

Evaluate: automated nudge system for at-risk learners (email, push notification, SMS), instructor alert dashboard, recommended interventions (additional resources, tutoring, schedule adjustment, peer support), intervention effectiveness tracking, intervention escalation (automated nudge failed, escalate to advisor contact), learner consent and transparency (do learners know they are being monitored and flagged).

============================================================ PHASE 6: ADAPTIVE SEQUENCING ============================================================

Step 6.1 -- Adaptive Algorithm

Evaluate: adaptation methodology (rule-based branching, item response theory, knowledge space theory, Bayesian knowledge tracing, reinforcement learning), adaptation granularity (course-level path vs. within-lesson content), adaptation triggers (assessment results, time-on-task, learner request, engagement patterns), adaptation speed (immediate after each interaction vs. end-of-module), learner control (can learners override adaptive recommendations).

Step 6.2 -- Content Sequencing

Evaluate: prerequisite knowledge verification before advancing, remediation content insertion when gaps detected, enrichment content for advanced learners, content format adaptation (video for visual learners, text for readers -- if learning style supported), pacing adjustment (slower for struggling, accelerated for proficient), spaced repetition integration for retention, interleaving practice across topics.

Step 6.3 -- Adaptation Effectiveness

Evaluate: whether adaptive learners outperform non-adaptive (A/B testing), completion rate impact of adaptive sequencing, time-to-mastery comparison, learner satisfaction with adaptive experience, edge cases (learner with inconsistent performance -- advanced in some areas, struggling in others), cold-start problem (how does the system adapt for a brand-new learner with no history).

============================================================ PHASE 7: CREDENTIAL STACKING ============================================================

Step 7.1 -- Credential Architecture

Evaluate: credential types available (micro-credential, certificate, certification, degree pathway credit), credential hierarchy (do micro-credentials stack into certificates, certificates into degrees), credit articulation (do credentials carry academic credit), prior credential recognition (can existing credentials reduce requirements), credential portability (recognized by other institutions and employers).

Step 7.2 -- Stacking Pathway Design

Evaluate: clear credential progression maps (what leads to what), multiple entry points based on prior learning, time and cost transparency per credential level, stacking flexibility (can learners stack from different providers), credential expiration and renewal requirements, cumulative transcript or record.

Step 7.3 -- Credential Value

Evaluate: employer recognition data for each credential, salary impact data, employment outcome tracking by credential level, credential vs. experience trade-off analysis, credential market saturation awareness (too many holders reduces value), credential alignment with industry hiring requirements.

============================================================ PHASE 8: ROI TRACKING ============================================================

Step 8.1 -- Cost Tracking

Evaluate: per-learner cost calculation (tuition, materials, technology, instructor time, support services), per-program cost tracking, cost per completion (accounting for dropouts), cost per credential issued, cost comparison across delivery modes (in-person, online, hybrid, self-paced).

Step 8.2 -- Outcome Tracking

Evaluate: employment outcome tracking (job placement rate, time to employment), wage gain tracking (pre-training vs. post-training earnings), employer satisfaction surveys, credential utilization rate (are credentials used in employment), career advancement tracking (promotions, role changes), longitudinal tracking duration (6 months, 1 year, 3 years post-completion).

Step 8.3 -- ROI Calculation

Evaluate: ROI methodology (individual ROI for learner, program ROI for institution, social ROI for public funders), ROI time horizon, discount rate application, comparison methodology (treatment vs. comparison group, before-after), ROI reporting for stakeholders (funders, workforce boards, employers), ROI by subpopulation (does the program work equally well for all learners).

Write analysis to docs/training-path-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 ============================================================

Training Path Analysis Complete

• Report: docs/training-path-analysis.md
• Prerequisite mapping quality factors: [count]
• Competency framework elements assessed: [count]
• Learning outcome measures reviewed: [count]
• Completion prediction model components: [count]
• Adaptive sequencing capabilities: [count]
• Credential stacking pathways evaluated: [count]
• ROI tracking metrics assessed: [count]

Critical findings:

1. [finding] -- [learner outcome impact]
2. [finding] -- [competency measurement concern]
3. [finding] -- [pathway design gap]

Top recommendations:

1. [recommendation] -- [expected improvement in completion rates]
2. [recommendation] -- [expected improvement in competency validation]
3. [recommendation] -- [expected improvement in employment outcomes]

NEXT STEPS:

• "Run /skill-gap to evaluate how training paths align with identified workforce gaps."
• "Run /employer-matching to assess whether training completers match employer requirements."
• "Run /student-personalization to analyze adaptive learning capabilities in detail."

DO NOT:

• Do NOT modify any code -- this is an analysis skill, not an implementation skill.
• Do NOT include real learner names, grades, or personally identifiable information in output.
• Do NOT evaluate training content quality -- evaluate the system's ability to sequence, track, and optimize training.
• Do NOT ignore equity in completion prediction -- models trained on historical data may encode systemic biases.
• Do NOT treat completion as the only success metric -- competency attainment and employment outcomes matter more.
• Do NOT overlook the cold-start problem in adaptive systems -- new learners need reasonable defaults before adaptation kicks in.
• Do NOT assume credential stacking is always beneficial -- credential inflation can devalue individual credentials.
• Do NOT calculate ROI without accounting for selection bias -- higher-performing individuals may self-select into training.

============================================================ 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:

### /training-path — {{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.