Rubber Duck
- Dominant language
- Shell
- Stars
- 11.2k
- Forks
- 1.9k
- Avg merge
- 14h 16m
- Merged PRs (30d)
- 6
Description
### Describe the bug
rubber-duck reviews sometimes use the same model family as the primary session instead of selecting an independent reviewer, reducing the value of adversarial review.
i have observed this on a variety of models. most recently using 5.6 Terra - Max
### Affected version
v1.0.78
### Steps to reproduce the behavior
Have code/plan/contract whilst working in a gpt-5.6-terra session and request a rubber-duck review...
you then see this: Rubber-duck(gpt-5.6-terra) Assess audit design quality
└ Agent started in background with agent_id: audit-design-duck. You'll be notif...
### Expected behavior
I would expect a different model family selected from the current model being used in the session.
### Additional context
_No response_
Contributor guide
Research direction
Start by reproducing the rubber-duck review in the reported gpt-5.6-terra session and trace the model-selection entry point used for the background audit-design-duck agent. Done means the reviewer selects a different model family from the primary session while preserving the review request behavior.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- shell
- Domain
- ai, cli
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 45/100