Commit Graph

374 Commits

Author SHA1 Message Date
Jorge Manrubia fa82dc0384 Remove code related to Fizzy Do /command support and filtering
https://3.basecamp.com/2914079/buckets/37331921/messages/8961776532
2025-08-18 13:41:20 +02:00
Stanko K.R. 4f219fadf3 Add test for the price calculation 2025-08-13 15:14:12 +02:00
Stanko K.R. aa11c82e1c Add test for the list comments tool 2025-08-13 15:14:12 +02:00
Stanko K.R. 0807337ede Add test for the list collections tool 2025-08-13 15:14:12 +02:00
Stanko K.R. 8574e53d83 Add test for the list users tool 2025-08-13 15:14:12 +02:00
Stanko K.R. 33daa866f3 Test the list cards tool 2025-08-13 15:14:12 +02:00
Stanko K.R. c1bd68d0e6 Add tests for OrderClause 2025-08-13 15:14:12 +02:00
Stanko K.R. 7f38586bda Fix crash when searching 2025-08-13 15:14:12 +02:00
Stanko K.R. 8a8888c3ef Tidy up #ask & #respond 2025-08-13 15:14:12 +02:00
Stanko K.R. db4415d6c7 Add tests for conversations and messages 2025-08-13 15:14:12 +02:00
Stanko K.R. 09739fdba4 Add tests for the Conversation object 2025-08-13 15:14:12 +02:00
Stanko K.R. 13eeb8574d Add the ask command 2025-08-13 15:14:12 +02:00
Jason Zimdars aa13853ae4 Add a third, "On deck" column in Collections 2025-08-12 17:58:12 -05:00
Jorge Manrubia 588f03e927 Remove insight experimental code
Also, downgrade LLM for commands to mini. It's much cheaper and I only upgraded because of the
interpretation complexities that the insight command introduced.

We are implementing a much better approach with Ask.
2025-08-11 13:40:56 +02:00
Mike Dalessio 9bdc273c76 Prefer Account#slug to Account.script_name
Caching the script name in a class instance variable seemed like a bug
to me.
2025-08-09 14:27:41 -04:00
Andy Smith ece1751538 Merge pull request #805 from basecamp/bold-event-actions
Update system message to highlight verbs
2025-07-28 13:23:34 -05:00
Andy Smith 07a470a8ac Update tests 2025-07-28 13:12:53 -05:00
Jorge Manrubia d90fac853f Add missing test 2025-07-25 10:09:00 +02:00
Jorge Manrubia 711d9207ed Little refactor + missing test 2025-07-25 10:04:07 +02:00
Jorge Manrubia d5bedd812d Regenerate test's fixtures to see if that fixes it for JZ 2025-07-25 06:27:12 +02:00
Andy Smith 8e795f183d Update system message to highlight verbs 2025-07-24 13:42:19 -05:00
Jorge Manrubia c128b07203 Add test 2025-07-24 17:47:42 +02:00
Jorge Manrubia 7de45cc061 Remove vcr_record! and fix VCR fixtures due to different timestamps 2025-07-24 16:48:06 +02:00
Jorge Manrubia b2ee79bbd0 Format 2025-07-24 16:48:06 +02:00
Jorge Manrubia 5ad2d09eae Move the hack to fetch completed cards to the insight command logic
We need to rework this as part of the filters revamp. "indexed_by" being
a mix of filters keeps biting us.
2025-07-24 16:48:05 +02:00
Jorge Manrubia fc26329daf Restore default to open cards. Use special "all" value instead.
The change could have side effects (e.g: when running commands).

This also include further changes.
2025-07-24 16:48:04 +02:00
Jorge Manrubia 018c746d2a Make the prompt smarter about the current context to make better calls 2025-07-24 16:48:03 +02:00
Jorge Manrubia 8bb7783215 Extract "prompt generation" logic to models for events, comments and cards
So that we can reuse in other scenarios
2025-07-24 16:48:03 +02:00
Jorge Manrubia dd28e3d1e9 Replace special value with just not asuming that filters are about open cards by default 2025-07-24 16:48:03 +02:00
Jorge Manrubia 9d0887f6bd Use a workaround with a special indexed_by filter to "query all" 2025-07-24 16:48:02 +02:00
Jorge Manrubia 83d743fe2a Make the translator use the /insight command 2025-07-24 16:48:02 +02:00
Jorge Manrubia 1c6adf54a8 Restore semantic searches 2025-07-24 16:47:57 +02:00
Jorge Manrubia 6080c2aa9b Refine when to filter by cards, make it more relaxes with the expressions it accepts 2025-07-24 16:47:53 +02:00
Jason Zimdars 84f19183be Add model test for read and unread 2025-07-23 17:28:15 -05:00
Jorge Manrubia fc1ff7bf5e Fix tests 2025-07-23 12:21:20 +02:00
Jorge Manrubia fd0484176c Format 2025-07-22 18:59:10 +02:00
Jorge Manrubia 70f19e54b5 Add assertion on HTML content 2025-07-22 18:47:48 +02:00
Jorge Manrubia e87ed26f75 Add tests 2025-07-22 18:45:07 +02:00
Jorge Manrubia 2cd4dd8709 Prevent recursion when invoking AI
This can result in stack overflow errors for missing commands.

https://fizzy.37signals.com/5986089/collections/7/cards/1135

Follow up to https://github.com/basecamp/fizzy/pull/766
2025-07-21 10:25:26 +02:00
Jorge Manrubia 39c7a54ad6 Add /user command
The LLM was relying on it. Having it missing resulted in a stack overflow error!

https://fizzy.37signals.com/5986089/collections/7/cards/1135
2025-07-18 17:59:02 +02:00
Mike Dalessio 15876b8762 Update Signup to use the queenbee_id for the tenant name
and not the subdomain. This change should have been in #665
2025-07-18 10:11:10 -04:00
Jorge Manrubia a1e3835eb8 Prevent 500 when running /search with empty string 2025-07-18 10:55:36 +02:00
Jorge Manrubia 44873e8ac4 More refinements 2025-07-18 06:50:13 +02:00
Jorge Manrubia d81568ff57 Filter by stage id, much better collection filtering 2025-07-18 05:50:22 +02:00
Jorge Manrubia b30bc85f4e Rework prompt to make it less ambiguous and better overall 2025-07-18 05:11:08 +02:00
Jorge Manrubia 5f04696fe8 Rewrite prompt to be more accurate and example-driven 2025-07-17 13:32:36 +02:00
Mike Dalessio 9f887e17f5 Destroy the SignalId::User when a user is deactivated 2025-07-15 18:30:23 -04:00
Jorge Manrubia 3167126bda Rename test 2025-07-04 20:03:59 +02:00
Jorge Manrubia 919211244b Review the code to move cards
- Scope accessed records
- Keep consistency by moving events too
2025-07-04 19:56:48 +02:00
Jorge Manrubia 3fb3e5b48e Generalize and delete mentions too 2025-07-04 19:08:29 +02:00