[codex] add Claude Code relay security guide

[toby-bridges]

Summary

• publish Claude Code, ANTHROPIC_BASE_URL, and Prompt Steganography: How to Audit AI API Relays
• connect the guide from the README, English Pages guide grid, Chinese Pages guide list, and sitemap
• add article/FAQ structured data plus social metadata for the new landing page

Evidence boundary

• attribute the reported Claude Code 2.1.196 behavior to Thereallo's June 2026 reverse-engineering analysis
• separate client-side prompt steganography from relay prompt injection
• state explicitly that API Relay Audit does not inspect the Claude Code binary, capture its raw system prompt, or certify a relay as safe

Validation

• git diff --check
• python3 -m pytest tests/test_static_site.py -v (6 passed)
• python3 scripts/collect-metrics.py --check
• desktop and 500px mobile renders checked with headless Chrome