Disable Claude Code's Auto Prompt Suggestions

Problem

Claude Code by default shows "Try..." suggestions after conversations, such as:

Try "write a test for <filepath>"
Try "fix lint errors"

If you find these suggestions distracting, you can disable them through configuration.

Claude Code uses GrowthBook as a feature toggle system. These configurations are stored in ~/.claude.json:

"cachedGrowthBookFeatures": {
  "tengu_willow_mode": "hint_v2",
  "tengu_prompt_suggestion": true
}

These values are remotely cached, periodically refreshed and overwritten by Anthropic's server. Direct modification is ineffective.

Add officially supported configuration in ~/.claude/settings.json:

{
  "promptSuggestionEnabled": false
}

This local configuration has higher priority than remotely delivered GrowthBook configuration.

Setting	Purpose
`promptSuggestionEnabled`	Disable all prompt suggestions in input box and after conversations
`tengu_willow_mode`	Controls "Try..." suggestions after conversations (gets remotely overwritten)
`tengu_prompt_suggestion`	Controls gray placeholder hints in input box (gets remotely overwritten)

After modification, restart Claude Code, suggestions should no longer appear.

cat ~/.claude/settings.json

tengu_* is Anthropic's internal codename, not Tencent Cloud configuration
GrowthBook is Anthropic's feature toggle service for dynamically controlling client features
If you want to restore suggestions later, change promptSuggestionEnabled to true or remove the configuration