perf(mothership): restore static markdown parse for settled chat messages

[waleedlatif1]

Summary

• Chat scroll got laggy after the mothership v0.8 merge (feat(mothership): mixture of models, search agent, persistent subagents, fork chat, inline questions (mothership v0.8) #5410), which removed mode='static' from the chat markdown renderer. Every message then rendered through Streamdown's streaming parser — remend + incomplete-markdown repair + per-block re-parse, running the rehype raw/sanitize/harden chain once per block.
• Because chat rows are virtualized, every up/down scroll remounts the messages crossing the overscan boundary and re-pays that N-block cost. That's the lag.
• Restore mode='static' for never-streamed mounts (reloaded history, or an in-session message scrolled out of the virtualized window and back): one whole-document parse instead of streaming's per-block re-parse. A completed message needs no incomplete-markdown repair, so static is the correct — and much cheaper — mode.
• In-session streaming is unchanged: it keeps the streaming parser for its mount life, so there's no drain re-parse flash.
• Cache Prism highlight output in a module-level bounded LRU (512, LRU eviction) so a code block re-highlights at most once across the unmount/remount virtualization does on scroll — a component useMemo would not survive the unmount.

Root cause

Verified by commit archaeology + reading streamdown@2.5.0's compiled parser: mode: 'static' renders the whole message in one ReactMarkdown pass and skips parseIncompleteMarkdown; mode: 'streaming' splits into N blocks and renders each as its own ReactMarkdown subtree, running the expensive rehype-raw chain per block.

Type of Change

• Bug fix (performance regression)

Testing

• tsc clean (0 errors), biome clean, 108/108 message-content tests pass.
• /cleanup (8 parallel quality passes: effects, state, memo, callback, react-query, url-state, emcn, comments) returned zero findings.

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
docs	Skipped		Jul 17, 2026 11:40pm

[cursor]

PR Summary

Low Risk
UI rendering and performance only; parser-mode splitting is nuanced but scoped to chat markdown display with no auth or data-path changes.

Overview
Fixes chat scroll lag after mothership v0.8 by no longer forcing every message through Streamdown’s expensive streaming parser on every virtualized remount.

Streamdown mode: Reloaded history (and any mount that never streamed this session) now use mode='static' — one whole-document parse instead of per-block rehype work. Messages that streamed in-session stay on the streaming parser for the mount’s life (parserTree), even after animation drains; only the animate/remount key (streamingTree) drops at drain so spans shed without re-parsing through static (which would flash on unbalanced markdown).

Syntax highlighting: Code blocks go through a module-scoped bounded LRU cache (512 entries) for Prism output, since row virtualization remounts rows and component useMemo would not survive.

^{Reviewed by Cursor Bugbot for commit 76892a3. Configure here.}

[greptile-apps]

Greptile Summary

This PR improves rendering performance for virtualized chat messages. The main changes are:

• Uses static markdown parsing for settled messages that never streamed during the current mount.
• Keeps the streaming parser active for messages streamed during the current mount.
• Adds a bounded module-level LRU cache for Prism highlight output.
• Skips caching when the requested Prism grammar is unavailable.

Confidence Score: 5/5

This looks safe to merge.

• The unavailable-grammar path now returns before cache insertion, so later grammar registration is not hidden by stale fallback output.
• Registered grammars use resolved language names in the cache key.
• The imported Prism helper matches the arguments used by the new wrapper.
• No blocking issue related to the previous highlighting failure remains.

Important Files Changed

Filename	Overview
apps/sim/app/workspace/[workspaceId]/home/components/message-content/components/chat-content/chat-content.tsx	Selects the markdown parser from the message lifecycle and caches registered-grammar Prism output across virtualized remounts.

_{Reviews (2): Last reviewed commit: "fix(mothership): don't cache fallback-hi..." | Re-trigger Greptile}

[waleedlatif1]

@greptile

[waleedlatif1]

@cursor review

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 76892a3. Configure here.}