What problem does it solve? Vague probability words like "possible" or "significant chance" are read as anything from 20% to 80% by different readers, and mixing confidence with likelihood in one sentence corrupts analytic assessments. This Skill converts numeric probabilities into calibrated terms from a declared standard and lints drafts for vague, hedged, or mixed-standard wording. ## Core Features & Use Cases - Probability-to-term conversion: Maps a numeric probability to the correct term in ODNI ICD 203, the UK PHIA Probability Yardstick, or the IPCC AR5 scale, handling shared boundaries, deliberate PHIA gaps, and nested IPCC bands. - Deterministic linting: The stdlib-only scripts/wep.py linter flags vague likelihood words, bare numbers without terms, numeric confidence, hedged terms, foreign-standard terms, and confidence-plus-likelihood sentences, with line:col findings and exit codes. - Confidence separation: Enforces the ICD 203 rule that confidence (high/moderate/low, with its sourcing basis) lives in its own sentence, never combined with a likelihood term. - Use Case: A board brief says "there is a significant chance the fab slips and we have high confidence it is likely." The Skill rewrites it as "There is a roughly even chance (45–55%) that the fab slips" plus a separate confidence sentence with its basis, and the linter confirms zero errors. ## Quick Start Use the estimative-language skill to rewrite my attached assessment using ICD 203 probability language and lint it for vague wording.
