Hi, when using the can i deploy GH action, failure...
# pactflow
p
Hi, when using the can i deploy GH action, failures are not verbose, is there an easy way to show exact failure reasons in a nice readable way in CI output?
m
Just so I’m clear, is it the truncation of columns that’s the problem?
The failure reasons should be in the “verification results” section, which links to the PactFlow UI
p
Yup right the link for that was getting truncated by Github fixing now, but would be cool if we could see the error formatted nicely in ci output like "this endpoint is being removed in this pr"
so my main issue is gone now, all g
👍 1
m
Ah right, yeah we don’t provide the specific error details here - this is really a query on the compatibility, but the idea is to link to the details for further information.
p
this is in the pr pipeline checks for us
👍 1
is there any way to get specific details in ci?
so that devs can direct their agent at ci logs instead of going into the ui
ik they could also use an api key but would be smoother
m
It’s a good question. We have APIs to get this information, of course: https://developer.smartbear.com/pactflow/default/pactflow_saas_api#/ Rather than spitting all that info into the can-i-deploy tool, we could potentially add more tools to it. Another thing we’re looking at is CI native contract testing agents. You’re using GH actions it seems, would something like that be helpful? e.g. Pull request goes up, CT agent reviews it, suggests changes, coverage reports etc.
p
hm, we try to avoid any extra agent reviews on PRs
we already have a pretty comprehensive internal one
m
For this use case, you could install our MCP and AI Skills, and ask it to diagnose the failure (https://support.smartbear.com/swagger/contract-testing/docs/en/user-guide/ai-skills.html) . It should be able to figure things out (see also the
/ai
page in your account)
gtg I have a call, but will pop back here later
p
i guess we could add a step that gets triggered on failure and then call sthe api for the exact failure
we just don't want devs to have to set up the MCP each , just get everything in ci output
m
i guess we could add a step that gets triggered on failure and then call sthe api for the exact failure
That seems like an opportunity for us to help improve the developer and agent experience.
we just don’t want devs to have to set up the MCP each , just get everything in ci output
makes sense, you want to reduce the friction on engineers having to manually set things up. cc @Joshua Ellis @Kevin Vaz as we think through our agent work.
👍 1
p
Cheers happy to any questions guys ^
🙏 1