fix: hide system reminders in chat

[lauridskern]

Summary

Prevent raw <system_reminder> XML from rendering in chat messages.

Context

Resolves #31. System reminders can appear in assistant-visible transcript text, but that internal markup should not be displayed directly to users.

Changes

• Sanitize assistant, reasoning, and status output transcript text while building chat thread items.
• Remove <system_reminder>...</system_reminder> blocks before messages reach the visible chat rows.
• Drop messages that contain only a system reminder after sanitization.
• Added regression coverage for both mixed-content and reminder-only assistant messages.

Testing

• cargo check --manifest-path src-tauri/Cargo.toml
• git diff --check

Attempted but unavailable in this environment:

• bun test src/components/chat/ChatWorkRow.test.ts (bun is not installed)
• npx tsc -b (npx is not installed)

Links

• Related issue: Prevent system_reminder XML from rendering in chat messages #31

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: fb94b6bfc3

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Avoid trimming normal assistant text in sanitizer

removeSystemReminderMarkup always applies .replace(/\n{3,}/, ...) and .trim() even when the message contains no <system_reminder> tag, so this commit now rewrites all assistant/reasoning/status output text rather than only removing internal markup. That can alter user-visible formatting (for example, leading indentation used for Markdown code blocks or intentional extra blank lines) in ordinary responses, which is a behavior regression unrelated to reminder removal.

Useful? React with 👍 / 👎.