Proposal
Implement a local, high-performance autocomplete (typeahead) mechanism for commit message generation using a Trie data structure.
Context
Instead of relying on real-time LLM API calls, which introduce latency and performance overhead in the TUI, a local Trie-based approach could provide instant suggestions for common commit message verbs, scopes, or frequently used phrases.
Implementation Ideas
- Non-blocking: The autocomplete should be local and instantaneous to ensure the TUI remains responsive.
- Trigger: Rather than live-as-you-type, it could be triggered via a specific keystroke (e.g.,
TAB or CTRL-SPACE) to suggest completions.
- Data Source: The Trie could be populated with common conventional commit structures, previous commit messages, or a project-specific word list.
This approach avoids the complexity and UX pitfalls of live LLM typeahead while still providing productivity gains.
Proposal
Implement a local, high-performance autocomplete (typeahead) mechanism for commit message generation using a Trie data structure.
Context
Instead of relying on real-time LLM API calls, which introduce latency and performance overhead in the TUI, a local Trie-based approach could provide instant suggestions for common commit message verbs, scopes, or frequently used phrases.
Implementation Ideas
TABorCTRL-SPACE) to suggest completions.This approach avoids the complexity and UX pitfalls of live LLM typeahead while still providing productivity gains.