| name | lmti-score |
| description | Score an LMTI-60 selftest result, compute dimension lean, dominant and secondary types, detect inconsistency from cross-reference clusters, and render the output as markdown, JSON, or HTML. |
LMTI Score
Use this skill when the user provides LMTI-60 answer output from lmti-selftest and wants the personality result.
Workflow
- Read
references/framework.json.
- Use
scripts/score_lmti.py to score the result.
- The input can be:
- raw JSON from
lmti-selftest
- markdown or prose that contains that JSON object
- a file path whose contents contain that JSON object
- Prefer running the script with the raw JSON on stdin when the result is pasted in the conversation.
- Use output format rules:
- default:
markdown
json when the user asks for machine-readable output
html when the user asks for rendered HTML
- By default, return the script output directly. Add extra explanation only if the user asks for it.
Notes
- The scorer reports all four dimension lean scores.
- It always returns one dominant type, but it also returns secondary type affinities because LMTI allows mixed tendencies.
- It computes a consistency score from the 20 three-question cross-reference clusters.
- It also flags straightlining and excessive neutral-answer bias.
- The scorer script supports
--format markdown|json|html.