Feature Request: context-aware autocomplete/suggestion functionality
Resolved 💬 1 comment Opened Aug 23, 2025 by mkusaka Closed Aug 23, 2025
What feature would you like to see?
I would like to request a context-aware autocomplete feature for the Codex CLI input field. This feature would provide intelligent, real-time suggestions for completing prompts based on the current conversation context.
Feature Description
When typing in the input field, the CLI should:
- Analyze the current conversation context (previous messages, code being discussed, files opened)
- Predict and suggest likely completions for the prompt being typed
- Display suggestions inline as ghost text (grayed out) after the cursor
- Allow Tab key to accept the suggestion
Examples
- Code-related context:
- After discussing a Python function, typing "add error h..." → suggests "add error handling to this function"
- After viewing a React component, typing "convert to..." → suggests "convert to TypeScript"
- Task continuation:
- After fixing a bug, typing "now run..." → suggests "now run the tests"
- After refactoring code, typing "update the..." → suggests "update the documentation"
- Pattern recognition:
- If user frequently asks "explain this code", typing "exp..." → suggests "explain this code"
- Learning from user's common phrase patterns and suggesting completions
Implementation Approach
- Use the conversation history and current context to generate predictions
- Could leverage the same LLM with a specialized prompt for completion
- Cache common patterns for faster response
- Make it configurable (on/off) in settings
Benefits
- Faster interaction: Reduce typing for common requests
- Better user experience: Similar to GitHub Copilot's inline suggestions
- Context preservation: Helps maintain focus on the current task
- Learning curve: Helps new users understand effective prompt patterns
Are you interested in implementing this feature?
No, but I'm happy to test and provide feedback.
Additional Information
This differs from simple command/file completion (which already exists) by providing intelligent, context-aware suggestions for the actual prompt content. It would make the Codex CLI feel more responsive and intelligent, similar to how modern IDEs provide code completion.
This issue has 1 comment on GitHub. Read the full discussion on GitHub ↗