Add AI Budget Tools, Enhanced Natural Language Queries, and Bug FixesF/raynab ai

[omarshahine]

Description

This PR introduces several new AI-powered tools and improvements for the Raynab extension.

It's a rather big PR since I had to modify some of the existing Forms and fix some bugs in them to support the add-transaction tool. I also took the liberty to add a budget check form to each command and send users to select the active budget. There were some places where the commands checked for an active budget but not all so I cleaned this up.

✨ New Features

AI Budget Tools:

• add-transaction: Create transactions with natural language and handoff to the Create Transaction form.
• get-budget: Query budget information, including age of money and monthly details.
• get-big-numbers: Track spending for today, this week, and this month.
• get-transactions: Search and filter transactions with natural language.
• get-categories: Query category information and spending.

💡 Example Queries

Transaction Management

• "@raynab add a transaction for $25.50 at Starbucks"
• "@raynab show me my recent transactions"
• "@raynab show me my Amazon purchases from last year"
• "@raynab what did I spend at Target in the past month?"
• "@raynab what was my transaction with Amazon on March 14"
• "@raynab show me my transactions with Amazon this month"
• "@raynab what did I spend at amazon last month?"

Account Queries

• "@raynab show me my account balances"
• "@raynab what is my checking account balance?"
• "@raynab show me my checking account"
• "@raynab how much is in my checking?"

Big Numbers

• "@raynab what are my big three numbers?"
• "@raynab show me my big numbers"

Budget Information

• "@raynab what is the age of money in my current budget?"
• "@raynab tell me about my this month's budget?"

💎 Improvements

Enhanced budget data retrieval and date formatting.
Improved error handling and user feedback for AI tools.
Support for natural language queries across all tools.

🐞 Bug Fixes

Fixed date format mismatch in budget month queries.
Fixed Validation issues on Create and Edit forms
Improved error handling in AI tool operations.

Checklist

• I read the extension guidelines
• I read the documentation about publishing
• I ran npm run build and tested this distribution build in Raycast
• I checked that files in the assets folder are used by the extension itself
• I checked that assets used in the README are located outside the metadata folder if they were not generated with our metadata tool

[raycastbot]

Thank you for your contribution! 🎉

🔔 @gogocharli you might want to have a look.

You can use this guide to learn how to check out the Pull Request locally in order to test it.

You can expect an initial review within five business days.

[greptile-apps]

PR Summary

This PR adds comprehensive AI-powered tools and natural language capabilities to the Raynab extension for managing YNAB budgets and transactions.

• Added 6 new AI tools in package.json for natural language budget management, including transaction creation, budget queries, and spending analytics
• Implemented consistent budget validation across commands using new checkForActiveBudget utility
• Added proper error handling and logging in AI tools with detailed debug information
• Added metadata folder for screenshots since there are view commands in package.json
• Improved transaction form validation and state management in both create and edit forms

Note: The PR should include a metadata folder with screenshots since there are view commands in package.json. Please refer to the Raycast Documentation for guidance.

_{26 file(s) reviewed, 31 comment(s)
Edit PR Review Bot Settings | Greptile}

[gogocharli]

Thanks for the superb PR @omarshahine! Reviewing this today 👍🏿

[andreaselia]

@gogocharli were you able to have a look at this?

[gogocharli]

There's one minor breakage in the TransactionEditForm but besides that everything is looking superb!

[omarshahine]

@gogocharli - all changes are done. Can you reply when you are ready with approved so the @raycastbot can merge

[gogocharli]

LGTM! Approved

[andreaselia]

Hey @omarshahine 👋

Looks like you have some failing evals, would you mind taking a look please?

You can re-run them locally using: npx @raycast/api@latest evals

[omarshahine]

Hey @omarshahine 👋

Looks like you have some failing evals, would you mind taking a look please?

You can re-run them locally using: npx @raycast/api@latest evals

I'm a bit confused and lost.

Despite the tests showing some failed, I am able to get them to work when I use them manually.

What model does the Evals use to do the tests? I'm using gpt-4.1-mini and it works.

If I use Ray-1 it does this for example, I don't consider this a failure if the model asks a question

[omarshahine]

@andreaselia figured it out. Will fix soon.

[omarshahine]

@andreaselia So I have fixed things, but oddly eval tests seem non deterministic. They fail in random ways and just now all passed. I did three runs, one passed and two failed. Not sure what to do about that.

[omarshahine]

Ok, I made more changes to make this more deterministic.

[andreaselia]

LGTM, thanks for updating the evals too!

[github-actions]

Published to the Raycast Store:
https://raycast.com/CodetaroMiura/raynab

[raycastbot]

🎉 🎉 🎉

We've rewarded your Raycast account with some credits. You will soon be able to exchange them for some swag.