---
name: scorpio-objective-evaluate
description: Score objective solution packages against the fixed rubric without editing packages or selecting a winner.
---

# Objective Package Evaluator

A bounded evaluation skill for the fourth stage of `scorpio objective`. It scores coherent solution packages against the fixed rubric and records uncertainty explicitly so downstream critique and synthesis can compare packages without conflating confidence with score.

## Purpose

This skill exists to evaluate packages, not to improve them.

It must:

1. score each solution package against every rubric dimension
2. preserve uncertainty notes and evidence quality
3. keep scores comparable across packages
4. leave package editing, critique, and recommendation to later skills

## Inputs

Read:

1. `{docs.planning}/OBJECTIVE-PARAMETERS.json`
2. `{docs.planning}/SOLUTION-PACKAGES.json`
3. `{docs.planning}/business-objective-workflow/SPEC-objective-scoring-rubric.md`

Do not depend on downstream critique or recommendation artifacts.

## Shared Rules

1. Score packages without editing packages.
2. Use the fixed global rubric from the planning pack.
3. Preserve evidence versus inference in every scorecard.
4. Capture uncertainty explicitly instead of smoothing it away.
5. Do not select a winner or remove weak packages.

## Scoring Contract

Use the fixed rubric dimensions and weights:

- `user_value`
- `business_viability`
- `distribution_realism`
- `retention_potential`
- `monetization_fit`
- `trust_safety_fit`
- `execution_feasibility`
- `speed_to_learning`
- `differentiation`
- `evidence_quality`

Each dimension uses the 1-10 scale defined in the rubric. Avoid false precision and do not inflate scores casually.

## Required Output

Write:

- `{docs.planning}/SOLUTION-SCORECARDS.json`

If the runtime provides a request-scoped objective directory, mirror the artifact there as well, but keep the root canonical file updated in the same run.

## Canonical Output Shape

`SOLUTION-SCORECARDS.json` should follow the schema contract:

```json
{
  "version": 1,
  "weights": {
    "user_value": 15,
    "business_viability": 15,
    "distribution_realism": 10,
    "retention_potential": 10,
    "monetization_fit": 10,
    "trust_safety_fit": 15,
    "execution_feasibility": 10,
    "speed_to_learning": 5,
    "differentiation": 5,
    "evidence_quality": 5
  },
  "scorecards": [
    {
      "package_id": "PKG-001",
      "scores": {
        "user_value": 0,
        "business_viability": 0,
        "distribution_realism": 0,
        "retention_potential": 0,
        "monetization_fit": 0,
        "trust_safety_fit": 0,
        "execution_feasibility": 0,
        "speed_to_learning": 0,
        "differentiation": 0,
        "evidence_quality": 0
      },
      "derived": {
        "composite_score": 0,
        "risk_adjusted_score": 0,
        "learning_value_score": 0
      },
      "confidence_level": "low | medium | high",
      "strongest_reason_to_build": "",
      "strongest_reason_not_to_build": "",
      "critical_assumption": "",
      "fatal_risk": "",
      "fastest_validation_step": "",
      "notes": {
        "known_evidence": [],
        "major_inferences": [],
        "uncertainty_notes": []
      }
    }
  ]
}
```

## Scorecard Requirements

Each scorecard must record:

1. per-dimension scores for every package
2. weighted rollups for comparative ranking
3. a confidence level distinct from the score itself
4. the strongest reason to build
5. the strongest reason not to build
6. the critical assumption that most affects the result
7. the fatal risk if the package is weak
8. the fastest validation step
9. uncertainty notes that preserve missing evidence or inference gaps

## What Not To Do

- Do not write `{docs.tasks}/**`.
- Do not edit `SOLUTION-PACKAGES.json`.
- Do not delete weak packages.
- Do not produce `SOLUTION-CRITIQUE.md`.
- Do not produce `RECOMMENDED-SOLUTION.md`.
- Do not reopen package selection.

## Handoff

When evaluation is complete, hand the scorecards to `scorpio-objective-critique` and `scorpio-objective-synthesize`.