feat(parser): render OpenCode task as Agent cards with nested tool calls

Transform OpenCode "task" tool calls with subagent_type into unified
Agent card display: rewrite as "Agent" tool with subagent_type, prompt,
description, and model; compute duration from time fields; query
sub-agent session parts from SQLite for nested tool call lists; extract
<task_result> content to strip preamble noise from agent output; guard
Streamdown code plugin against unsupported Shiki language identifiers
(e.g. "##", "function") in fenced code blocks from tool output.

src/components/ai-elements/message.tsx

@@ -327,7 +327,22 @@ export const MessageBranchPage = ({
 export type MessageResponseProps = ComponentProps<typeof Streamdown>
 
 const math = createMathPlugin({ singleDollarTextMath: true })
-const streamdownPlugins = { cjk, code, math, mermaid }
+
+// Wrap the code plugin to guard against unsupported language identifiers
+// (e.g. "##", "function") that appear in fenced code blocks from tool output.
+// Without this, Shiki's createHighlighter tries to load unknown grammars and
+// produces noisy console errors.
+const safeCode: typeof code = {
+  ...code,
+  highlight(options, callback) {
+    const lang = code.supportsLanguage(options.language)
+      ? options.language
+      : ("text" as typeof options.language)
+    return code.highlight({ ...options, language: lang }, callback)
+  },
+}
+
+const streamdownPlugins = { cjk, code: safeCode, math, mermaid }
 
 // remark-math only supports `$` delimiters. Convert LaTeX-style
 // `\[...\]` / `\(...\)` to `$$...$$` / `$...$` so they are recognized.