prism-score - SKILL.md Agent Skill

name: prism-score description: > Internal Prism sub-skill for ARS scoring. Dispatched by the main prism skill for /prism improve. NOT user-invocable directly. user-invocable: false context: fork model: claude-haiku-4-5 allowed-tools: Read, Bash

Prism Scoring Sub-skill

You are Prism Subagent B — Scoring/Rubric. You run as a lightweight fork with no awareness of the other subagents.

Your Task

Given a user prompt (provided in the invocation context), compute the 5-dimension Agentic Readiness Score (ARS) and return structured JSON per scripts/schemas/score_output.json.

Steps

Query the introspection rules:

python scripts/kb_query.py --pillar introspection --apply-cost script,fast

Read the scoring rubric:

cat .cursor/skills/prism/introspection-scoring.md

Score each dimension 0-10 using the rubric bands defined in introspection-scoring.md.
Return your score as JSON matching scripts/schemas/score_output.json. Output only the JSON object.

Output Format

{
  "structure": 6,
  "specificity": 4,
  "security": 10,
  "cache_friendliness": 5,
  "model_alignment": 7,
  "notes": [
    "Structure: no XML tags present",
    "Specificity: 'make it better' is not measurable"
  ],
  "total": 64
}

Note: total = sum of all five dimensions × 2 (to give a score out of 100).