Global chat job code#495
Merged
Merged
Conversation
|
|
||
| # notes | ||
|
|
||
| Reproduces a real session in which the assistant returned only "..." after ~19s. User is on the update-mailchimp step (adaptor `@openfn/language-mailchimp@latest`) of a multi-step workflow that also has Google Sheets and Gmail steps. Prior conversation generated the Mailchimp campaign code and explained the spread operator. The current user turn is an open-ended teach-me request that names "the hardest function for this adaptor" — phrasing is ambiguous and the model's behaviour here has historically been unreliable. |
Collaborator
There was a problem hiding this comment.
Definitely think that acceptance tests isn't the place for this sort of thing. It's an integration test.
But it's ok for now since we don't have integration tests . We'll review later
|
|
||
| ## content | ||
|
|
||
| What does the cursor() call at the top of this job actually do? I didn't write it and I'm not sure why it's there. |
Collaborator
There was a problem hiding this comment.
when I'm reviewing these tests it often feels useful to have the question be the first thing in the file
|
|
||
| # notes | ||
|
|
||
| Multi-turn conversation entirely about the same job code. The user has already asked two clarifying questions about the upsert step in a Salesforce contact sync and now asks a follow-up that requires keeping prior context in mind ("you mentioned external IDs — show me how to pick one"). The router should send this to job_code_agent. The response should pick up the thread from the prior assistant message rather than restarting the explanation. Inline code snippets in chat text are fine; the thing to avoid is generating a `code_edits` / `job_code` attachment that renders as an intrusive diff in the editor. |
Collaborator
There was a problem hiding this comment.
the thing to avoid is generating a
code_edits/job_codeattachment that renders as an intrusive diff in the editor.
is that right? The user has asked for an edit - why are we telling the model not to send an attachment?
I ask because this test is failing locally for me
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Short Description
Testing improvements and bug fixes for a set of user scenarios targeting job code edits relating to just the step that the user is viewing (i.e. not several steps, not another step that the user is not viewing, and not the workflow structure).
Fixes #490
Implementation Details
New acceptance specs for global_chat
Architecture / planning docs
Bug fix in planner job-code stitching
edits were lost. Now captures the matched key and uses it for stitching. Affected both single and parallel job-code-tool execution paths.
Prompt fixes
AI Usage
Please disclose how you've used AI in this work (it's cool, we just want to know!):
You can read more details in our Responsible AI Policy