Claude Code: Emoji Shortcode Autocomplete in the Prompt
Claude CodeClaude Code now autocompletes emoji shortcodes directly in the prompt input. Typing :heart: inserts an emoji, and a partial shortcode like :hea surfaces matching suggestions. The feature is on by default and can be turned off with the emojiCompletionEnabled setting. It is a small but broadly noticeable quality-of-life change that touches every user typing into the CLI.
Key Takeaways
- Emoji shortcode autocomplete lets developers type
:heart:or a partial code like:heato insert or search for emoji directly in the prompt. - The feature ships enabled by default but can be turned off entirely via the
emojiCompletionEnabledsetting for teams that don't want it. - It mirrors a pattern familiar from Slack and GitHub, lowering the friction of adding emoji in a plain-text terminal input.
- The same release added transcript-write failure warnings, surfacing silent data-loss risks like a full disk or an inherited environment variable disabling session saves.
- This is a purely additive UX change with no workflow migration required, it changes nothing for users who never use emoji shortcodes.
- It reflects Anthropic's continued investment in polishing the terminal-native interaction model rather than only shipping agentic capability upgrades.
Sources & Mentions
2 external resources covering this update
A Small Typing Convenience, Everywhere
Anthropic added emoji shortcode autocomplete to Claude Code's prompt input in version 2.1.217. Developers can type a colon-wrapped shortcode such as :heart: and Claude Code will insert the corresponding emoji directly into the prompt. Typing a partial shortcode, like :hea, brings up a suggestion list to complete from, matching the pattern familiar from Slack, Discord, and GitHub comment boxes.
Why It's There
Claude Code's prompt box is a plain-text terminal input, which historically meant emoji had to be typed manually via OS-level pickers or pasted in from elsewhere. Shortcode completion removes that friction for anyone who wants to add tone, flag a note, or annotate a message quickly while working in the CLI.
Opting Out
The feature is enabled by default. Teams or individuals who find it intrusive, or who rely on literal colon-wrapped text in prompts, can disable it via the emojiCompletionEnabled setting.
Shipped Alongside
The same release, 2.1.217, also added warnings when transcript writes are failing (for example, due to a full disk) or when session saving is silently disabled by an inherited environment variable, replacing what was previously a silent data-loss risk with a visible notice.