add frontend-design rule, fix codex-reviewer, route table update

- add .claude/rules/frontend.md: auto-loads /frontend-design for ui/** files
- add frontend-design to CLAUDE.md route table
- fix codex reviewer: use direct Bash instead of subagent (permissions issue)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

.claude/rules/frontend.md

@@ -0,0 +1,6 @@
+---
+paths:
+  - "ui/**"
+---
+
+When working on frontend code, always use the `/frontend-design` skill for component design, layout decisions, and UI implementation.

.claude/skills/ax/SKILL.md

@@ -108,14 +108,13 @@ For each:
 
 ## Parallel Codex Review
 
-On every AX audit invocation, **immediately** spawn the `codex-reviewer` subagent in the background before starting your own audit:
+On every AX audit invocation, **immediately** launch a background codex review before starting your own audit:
 
-```
+```bash
-Agent(subagent_type="codex-reviewer", run_in_background=true,
+codex exec -c 'reasoning_effort="high"' "AX audit: $ARGUMENTS. Read CLAUDE.md, .claude/ directory, and config files. Find: missing docs, unclear commands, split-brain config, stale references. File paths and exact fixes. Do NOT spawn sub-agents. Answer directly in bullet points." 2>&1
-  prompt="AX audit: $ARGUMENTS. Read CLAUDE.md, .claude/ directory, and config files. Find: missing docs, unclear commands, split-brain config, stale references. File paths and exact fixes.")
 ```
 
-Continue your own audit without waiting. When the codex-reviewer returns, integrate its findings into Phase 3 (REPORT):
+Run this via `Bash` tool with `run_in_background: true`. Continue your own audit without waiting. When the codex output returns, integrate its findings into Phase 3 (REPORT):
 - Codex findings that match yours → strengthen confidence
 - Codex findings you missed → add to recommendations
 - Disagreements → address explicitly in the report

.claude/skills/critic/SKILL.md

@@ -58,14 +58,13 @@ Optional prose narrative follows after a blank line.
 
 ## Parallel Codex Review
 
-On every critic invocation, **immediately** spawn the `codex-reviewer` subagent in the background before starting your own analysis:
+On every critic invocation, **immediately** launch a background codex review before starting your own analysis:
 
-```
+```bash
-Agent(subagent_type="codex-reviewer", run_in_background=true,
+codex exec -c 'reasoning_effort="high"' "Critique: $ARGUMENTS. Read all relevant files. What will break? What's missing? What's over-engineered? Do NOT spawn sub-agents. Answer directly in bullet points." 2>&1
-  prompt="Critique: $ARGUMENTS. Read all relevant files. What will break? What's missing? What's over-engineered?")
 ```
 
-Continue your own critique without waiting. When the codex-reviewer returns, integrate its findings into your verdict:
+Run this via `Bash` tool with `run_in_background: true`. Continue your own critique without waiting. When the codex output returns, integrate its findings into your verdict:
 - Codex issues you missed → add to `breakpoints`
 - Codex agrees with you → note in `survives` as independent confirmation
 - Codex disagrees → address in prose narrative

CLAUDE.md

@@ -39,6 +39,7 @@ This file provides guidance to Claude Code (claude.ai/code) when working with co
 | GCP commands | `/gcloud` skill |
 | Stress-test a decision | `/critic` skill |
 | Audit agent config quality | `/ax` skill |
+| Frontend/UI work | `/frontend-design` skill (auto-loaded for `ui/**` files) |
 
 ## Git (Gitea)