implement-suggestion

# Implement Suggestion Take reviewer suggestions on one or more pull requests, validate each through adversarial review (`/critical`) and a confidence gate (`/confidence`), then hand off a structured **suggestion-pack** to a worker subagent that applies the approved changes inside the PR's worktree and pushes them — without opening a new PR. This skill is a **thin orchestrator**. The heavy reasoning lives in `/critical` and `/confidence`. Per-PR worktree isolation comes from `gw`. Plan authoring for architectural changes is delegated to `aw-planner`. The mechanical apply / commit / push runs inside a dispatched worker. > **Source of truth.** This `SKILL.md` is a thin index. Detailed procedures live > in `rules/*.md`, literal artefacts in `templates/*.md`. Load only what the > current phase asks for. ## Mode Detection Parse `$ARGUMENTS` once, in this order. First match wins. | # | Signal in `$ARGUMENTS` | Mode | | - | ------------------------------------------------------------------------------------------ | -------------------------- | | 1 | One or more `github.com/<owner>/<repo>/pull/<n>` URLs (with or without `#discussion_r…`) | **multi-pr** | | 2 | One bare PR number (`#123`) **and** the current directory is a PR worktree | **multi-pr** (n=1) | | 3 | Free-text prose, pasted comment body, or non-PR URL ...