Problem

Agent replies are full of code, and the chat renderers undersell them:

• iOS/shared chat UI renders via AttributedString(markdown:), which has no code syntax highlighting and no table support — GFM tables come out as raw pipes.
• Android already renders GFM tables/strikethrough/task lists via CommonMark, but fenced code blocks are plain monospace with no highlighting.

Proposal

• iOS/shared (OpenClawChatUI, also benefits the macOS app): add fenced-code syntax highlighting and GFM table rendering while keeping streaming render performance. Any new dependency needs a license entry per app policy.
• Android (ChatMarkdown.kt): add syntax highlighting for fenced code blocks (small library after a health check, or a minimal in-repo lexer for the common languages).

Impact

Code-heavy agent conversations become readable at a glance on both platforms; iOS additionally gains tables.