docs: require impeccable skill for design and frontend

Adds the design-system issue, scaffolds .impeccable/live/config.json against
web/index.html, and records the aoe spawn commands for the codex GPT-5.6
models in the driver table.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

CLAUDE.md

@@ -42,6 +42,27 @@ request that adds a heavy dependency to either is wrong.
 - No `unwrap()` in non-test code. The lint is on.
 - SQL goes through `sqlx` compile-time-checked queries.
 
+## Design and frontend work
+
+**Any issue labelled `area/web`, and any work that decides how something looks
+or behaves on screen, goes through the `impeccable` skill.** That includes the
+design system itself, component work, layout, visual hierarchy, empty and error
+states, and any change to the UI described in `DESIGN.md` §9.2 and §9.3. Invoke
+it before writing markup, not as a review pass afterwards.
+
+`.impeccable/design.json` is the token system — colours, typography, spacing.
+It is authored once by that skill and then treated as the source of truth;
+components consume tokens rather than literal values. `.impeccable/live/config.json`
+points the skill's live browser iteration at `web/index.html`.
+
+`~/tea/arcada` has the same setup and is worth reading for the shape of a
+finished `design.json`.
+
+This matters more here than in a typical CRUD app because the manual-search
+view (§9.3) is the whole reason for the project's UI existing — Radarr's is
+unusable specifically because of a layout decision. Getting it right is a design
+problem, not a markup problem.
+
 ## Working an issue
 
 1. Read `DESIGN.md`, then the issue, then the sections the issue cites.
@@ -110,19 +131,28 @@ alone. The loop:
 
 ### Model selection
 
-| Difficulty | Model |
-|---|---|
-| `difficulty/hard` | Opus 5 |
-| `difficulty/moderate` | Sonnet 5 |
-| `difficulty/easy` | Sonnet 5 |
-| `difficulty/trivial` | Haiku 4.5 |
+Sessions are spawned with `aoe`, which can run `claude`, `codex`, `opencode` and
+`pi`. The GPT-5.6 models are `terra` and `sol`, reached through the `codex` tool.
+
+| Difficulty | Model | Spawn |
+|---|---|---|
+| `difficulty/hard` | Opus 5 | `aoe add <path> --tool claude --model claude-opus-5 -l` |
+| `difficulty/moderate` | Sonnet 5 or `gpt-5.6-sol` | `aoe add <path> --tool codex --extra-args "-m gpt-5.6-sol" -l` |
+| `difficulty/easy` | `gpt-5.6-terra` | `aoe add <path> --tool codex --extra-args "-m gpt-5.6-terra" -l` |
+| `difficulty/trivial` | Haiku 4.5 | `aoe add <path> --tool claude --model claude-haiku-4-5-20251001 -l` |
+
+Use `-w <branch> -b` to put each session in its own git worktree, which is what
+makes parallel issues safe.
+
+`gpt-5.6-terra` is confirmed — it is the default in `~/.codex/config.toml`.
+`gpt-5.6-sol` is inferred from the sibling naming and should be verified on
+first use.
 
 Escalate one tier if a session fails CI twice on the same issue. Never
 de-escalate mid-issue.
 
-> The operator mentioned a model named "terra" for the lower tiers. No such
-> model was identified — confirm what it refers to and correct this table
-> before relying on it.
+`area/web` issues ignore this table's lower tiers — see the design section
+above. They go through the `impeccable` skill regardless of difficulty label.
 
 ### What a driver must not do