Menu
Internal Prism sub-skill for ARS scoring. Dispatched by the main prism skill for /prism improve. NOT user-invocable directly.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Optimize, sanitize, and score prompts using the three-pillar Prism methodology (Refraction, Sanitization, Introspection). Use when the user runs /prism hello, /prism improve, /prism sanitize, /prism score, /prism explain, /prism format, /prism hook [on/off/status — defaults to on], /prism patterns, /prism usage, /prism configure, or asks to optimize or analyze a prompt for an AI model.
Optimize, sanitize, and score prompts using the three-pillar Prism methodology (Refraction, Sanitization, Introspection). Use when the user runs /prism hello, /prism improve, /prism sanitize, /prism score, /prism explain, /prism format, /prism hook [on/off/status — defaults to on], /prism patterns, /prism usage, /prism configure, or asks to optimize or analyze a prompt for an AI model.
Internal Prism sub-skill for refraction planning. Dispatched by the main prism skill for /prism improve. NOT user-invocable directly.
Internal Prism sub-skill for sanitization analysis. Dispatched by the main prism skill for /prism improve. NOT user-invocable directly.
Based on SOC occupation classification
| 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 |
You are Prism Subagent B — Scoring/Rubric. You run as a lightweight fork with no awareness of the other subagents.
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.
python scripts/kb_query.py --pillar introspection --apply-cost script,fast
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.
{
"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).