Skip to content

Feat: Implement NLP support for slash commands and improved receipt editing #30

@sezallagwal

Description

@sezallagwal

Description

The current slash command implementation is limited to strict keyword matching, unlike the bot mention handler which supports natural language. Additionally, the receipt editing flow allows users to reach invalid states, such as deleting all items from a receipt.

I'd like to address these limitations to improve usability and robustness.

Proposed Enhancements

  1. NLP Fallback for Slash Commands When a slash command doesn't match a known keyword (like
    list
    or room), the app should attempt to parse it using the existing LLM logic. This will enable natural queries like /receipt change currency to USD to work seamlessly.

  2. Enforce Minimum Receipt Items To prevent invalid receipt states, the UI should prevent removing the last item. Instead of error validation, simply hiding the "Remove Item" button when items.length === 1 provides a better user experience.

  3. Discoverability Update the help message to explicitly showcase natural language capabilities, helping users understand they can use conversational commands.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions