githubnext / githubnext/ado-aw
[agent-issue]: Allow model-selected reviewers in create-pull-request
- Dominant language
- Rust
- Stars
- 23
- Forks
- 8
- Avg merge
- 4d 9h
- Merged PRs (30d)
- 22
Description
### Submission requirements
- [x] I generated this issue with an agent that used `.github/agents/ado-aw.agent.md`.
- [x] I reviewed the generated issue and confirm it is being filed directly in `githubnext/ado-aw`.
### Problem summary
`create-pull-request` supports reviewers only through static workflow frontmatter:
```yaml
safe-outputs:
create-pull-request:
reviewers:
- user@example.com
```
The model-facing `CreatePrParams` accepts title, description, repository, and labels, but not reviewers. This prevents workflows from selecting reviewers dynamically—for example, adding the assignee of the issue that triggered an autofix workflow.
### Reproduction details
1. Create an issue-triggered workflow that opens an Azure DevOps PR.
2. Provide the authoritative issue assignee to the model.
3. Ask the model to add that assignee as a reviewer.
4. The model cannot include reviewers in its `create-pull-request` call.
5. `update-pr` supports dynamic reviewers, but requires a numeric PR ID that does not exist until safe-output execution.
This leaves custom post-processing scripts or hardcoded reviewers as the only options.
Expected: the model can propose reviewers as part of `create-pull-request`, subject to safe-output policy and threat detection.
### Proposed next step
Add an optional `reviewers` field to `CreatePrParams`.
During safe-output execution, merge and deduplicate model-supplied reviewers with statically configured reviewers, then use the existing reviewer identity resolution and PR reviewer APIs.
Consider requiring explicit operator opt-in, such as `allow-agent-reviewers: true` or an `allowed-reviewers` policy. Update the tool schema, documentation, and executor tests accordingly.
Posted by Copilot CLI assistant.
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.