lineup-freshness-sweeplisted

# Skill: Lineup Freshness Sweep The `cross-tool-model-lineup-2026.md` knowledge bank is Tier-4 (fast-churn): model pickers, pricing, and context-window limits change weekly to monthly. A stale lineup produces confidently wrong recommendations. This skill produces a structured refresh checklist that a researcher (or `ravenclaude-core/deep-researcher`) can execute against primary sources. ## When to run this skill - The knowledge bank's `Last-retrieved` header is more than 4 weeks old. - A developer reports that a model described in the lineup is missing from their picker. - A developer hits an unexpected charge suggesting a retirement redirect is not documented. - The `researcher-reminder.yml` weekly sweep fires. ## Step 1 — Identify all dated claims in the knowledge bank Open `../../knowledge/cross-tool-model-lineup-2026.md` and catalogue every claim that carries a `[verify-at-use — YYYY-MM]` marker or a `Last-retrieved:` tag. Group by ecosystem: ``` GitHub Copilot section: - Model picker contents (all surfaces × all plans) - Plan-gated model list (Free sub-list, Pro, Business, Enterprise) - Org model rules feature availability OpenAI Codex section: - Default model id - Frontier / Pro model id - Reasoning-level options (flag names, levels available) - Spark / lightweight tier xAI Grok section: - Active model ids (flagship, fast, lightweight) - Retirement / redirect table - Any pricing-tier markers ``` ## Step 2 — Score each claim by churn risk |