Reproducing agent implementation

[TomasKorbar]

Description

Reproducing the issue is integral part of investigating or fixing any moderately complex issue or vulnerability. Currently we do not have any automation creating reproducers and we would benefit both during triage step and backport creation from having it.

Steps of the agentic workflow:

1. Get Jira issue, check the package exists
2. Get maintainer rules
3. Meditate on facts about the issue, investigate links, clone downstream repository branch according to affected version from JIRA
4. Investigate related parts of the code
5. Reserve machine from testing farm with appropriate release
6. Clone https://gitlab.com/redhat/rhel/tests, add directory where agent will be working on reproducer (test directory)
7. Copy the reproducer to the testing-farm machine and verify whether the issue was reproduced, if not edit the test and try again
   <---- from this point reproducer is working
8. return machine to testing farm
9. Create MR for https://gitlab.com/redhat/rhel/tests
10. Add private comment about it to the JIRA issue

vmihalko should be the one implementing this.

Benefit

No response

Importance

No response

Workaround

• There is an existing workaround that can be used until this feature is implemented.

Participation

• I am willing to submit a pull request for this issue. (Packit team is happy to help!)

[TomasTomecek]

Makes perfect sense to me. We could even reuse parts of the initial analysis from the triage and applicability agents that already do that.

[martinky82]

yep, agreed. Test creation naturally belongs to the triage, hence it makes very much sense to use triage agent's output for that.

[martinky82]

I reckon the agent should be able to recognize situations when the bug is in code but is not reproducible (I've met such cases) and document the conclusion then alongside with (proposal to) setting the Test Coverage to 'Regression Only'.