Current Scoring Surfaces

SurfaceDimensionsOutputSource
Communication analyticsprofessionalism 20%, clarity 20%, responsiveness 15%, emotional intelligence 15%, collaboration 15%, adaptability 15%overall_score 0-100src/core/analytics/scoring.py, src/core/analytics/user_analytics.py
Task evaluationaccuracy 30%, communication 20%, compliance 20%, timeliness 15%, initiative 15%; optional workplace behaviorpass, partial, fail; overall_score 0-100src/core/tasks/task_evaluation_service.py
Workplace behaviorcommunication quality, stakeholder management, ownership followthrough, meeting followthrough, relationship impact, recovery behavior, business judgmentpass, partial, fail; rubric scores 0-100src/core/simulation/context_frame.py
Competency snapshotscommunication clarity, stakeholder management, prioritization quality, risk handling, collaboration maturityreadiness level: developing, progressing, ready, advancedsrc/core/tasks/task_assessment_helpers.py
Session assessmentcommunication quality, decision quality, stakeholder outcome quality, delivery effectivenessrecommendation: keep momentum, focused recovery, immediate interventionsrc/core/tasks/task_assessment_helpers.py

Draft Competency Rubric

Use this as product-facing rubric language. It preserves current backend dimensions while making scoring explainable to learners and admins.
CompetencyWeightDevelopingProgressingReadyAdvancedEvidence
Communication clarity20%Messages or artifacts are vague, missing decisions, or hard to act on.Intent is mostly clear, but key owner/date/risk details are inconsistent.Clear, concise, audience-appropriate updates with needed context.Anticipates ambiguity, summarizes tradeoffs, and makes next actions obvious.Message quality, task artifact feedback, context-window quotes
Stakeholder management20%Misses stakeholders, uses wrong channel, or creates avoidable friction.Responds to stakeholders but needs reminders or uneven tone.Aligns right people, handles sensitive topics in right venue, follows up.Proactively manages expectations, escalates early, protects trust.Relationship edges, approvals, escalation events, thread/DM choices
Prioritization quality15%Works on wrong item, ignores urgency, or lets blockers linger.Recognizes priority but struggles under competing work.Sequences work by impact, deadline, dependency, and risk.Re-plans under pressure and communicates priority shifts cleanly.Task state, priority shifts, active task load, overdue/blocked signals
Risk handling15%Fails to spot compliance, customer, HR, or delivery risk.Identifies some risks but response is late or incomplete.Names material risks, chooses safe process, and documents mitigations.Turns weak signals into prevention steps and high-confidence escalation.Consequence events, sensitive-policy routing, policy grounding
Collaboration maturity15%Creates noise, ignores meetings/action items, or over-tags colleagues.Participates but needs nudges for follow-through and focus.Collaborates with clear asks, closes action items, and respects attention.Coordinates groups with minimal churn and strong shared ownership.Meeting lifecycle, action items, mention nudges, channel behavior
Delivery effectiveness15%Output misses criteria or creates rework.Partial output meets some criteria but needs revision.Output meets rubric and operational context.Output is high quality, grounded, timely, and reusable by stakeholders.Task evaluation, artifact review, session assessment

Score Bands

ScoreLevelMeaning
0-61DevelopingCoaching needed before higher-risk work.
62-71ProgressingBasic behavior improving; still needs targeted support.
72-81ReadyCan handle normal workplace tasks with expected oversight.
82-100AdvancedHandles ambiguity, risk, and stakeholders with strong judgment.
These bands match competency_readiness_level() thresholds today: developing under 62, progressing 62-71, ready 72-81, advanced 82+.

Draft Rollup Model

Draw.io source: documentations/architecture/JobSim_Simple_Diagrams.drawio

Implementation Notes

  • Keep backend source of truth unchanged at first: CompetencySnapshot stays five-dimension, SessionAssessment stays four-dimension.
  • Map draft delivery_effectiveness to session assessment directly and to task evaluation overall_score.
  • Store rubric explanations in existing metadata fields before adding schema, unless reporting needs first-class query support.
  • Use evidence_snippets, relationship_edges, consequence_events, and evaluation_context_windows for auditability.
  • Avoid inflating scores when data is missing. Current defaults start around 60 for competency and 70 confidence for session assessment, which is safer than auto-perfect scoring for workplace behavior.