core-reviewlisted

<!-- Note on the `model: haiku` pin above: - **Claude Code** honours it per-turn — this is why it exists (cheap-tier default so this pass, which runs at least once per PR + once per review round, does not multiply the cost of every autonomous cycle). - **VS Code Copilot's chat-customizations-evaluations linter** flags `model:` as an unsupported skill attribute (its allow-list for skills is: `argument-hint`, `compatibility`, `context`, `description`, `disable-model-invocation`, `license`, `metadata`, `name`, `user-invocable`). That warning in the Problems panel is **expected and cosmetic** — Copilot's runtime silently ignores unknown skill attributes, and skills have no independent `model:` boundary on Copilot anyway (they inherit the invoking prompt's model). Suppress-by-editing is not worth the split-ship complexity: keep the pin so Claude Code stays cheap. - **Codex** has no per-skill `model:` mechanism at all, so this line is inert there. --> # Silver Assist — Core Review (Pre-Review) A **pre-emptive consistency review** that runs *before* a reviewer (Copilot or a human) ever sees the branch, over a file set the caller scopes with `--budget` (diff → one-hop neighbours → whole repo; see the budget table below). It catches the classes of issues that trigger multi-round review loops — doc↔code drift, invalid code examples, broken links, stale indexes — so they are fixed in the first push instead of round 5. This skill is the revi