Address Holistic Review Workflow Feedback

[jeffhandley]

This draft collects and addresses feedback from the recently enabled Holistic Review automation. Maintainers are encouraged to push refinements directly to this draft while reviewing its results.

Holistic Review feedback addressed

• Calibrates complexity-versus-maintainability guidance and restores concise, evidence-focused review output.
• Centralizes per-PR orchestration state in a dedicated issue rather than adding workflow noise to reviewed PRs.
• Orders state creation before worker dispatch, recovers callback outcomes, and suppresses no-op report issues.
• Posts PR reviews only for new actionable feedback; routine LGTM, no-findings, and unchanged assessments complete without adding a review or being retried.

Other feedback addressed in merged PRs

• Update code review skill guidelines for suggestions formatting #131035

Note

This PR description was generated by GitHub Copilot.

[azure-pipelines]

Azure Pipelines:
Successfully started running 1 pipeline(s).
15 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[dotnet-policy-service]

Tagging subscribers to this area: @dotnet/runtime-infrastructure
See info in area-owners.md if you want to be subscribed.

[Copilot]

Pull request overview

This PR refines the Holistic Review workflow guidance to more explicitly evaluate when additional complexity in a PR is justified by its goals, and updates the compiled lockfile accordingly.

Changes:

• Add explicit complexity-vs-maintainability tradeoff assessment guidance (including when to escalate to ⚠️ Needs Human Review) in the Holistic Review instructions.
• Regenerate the holistic-review.lock.yml metadata to reflect the updated workflow source.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
.github/workflows/holistic-review.md	Adds guidance instructing reviewers to explicitly assess whether added complexity is proportionate and well-justified, and when to escalate for maintainer judgment.
.github/workflows/holistic-review.lock.yml	Updates the generated lock metadata (hash) corresponding to the workflow text change.

[MihaZupan]

Is there consideration for having it not post a comment at all if it's deemed not actionable by the LLM itself? Or would you still want it to always post, but perhaps with less verbosity?

E.g. in #130802 (review) there's a whole page of text for what boils down to Detailed Findings: None.
I also don't see much value in having it restate what the PR is doing, assuming that it matches what the description says, because the Copilot reviewer already posts a similar summary as a comment anyway, e.g. #130802 (review)

[Copilot]

Pull request overview

Copilot reviewed 9 out of 9 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 9 out of 9 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (1)

.github/workflows/holistic-review-completed.yml:62

• The workflow_call inputs are defined as pr_number and head_sha, but this job reads github.event.inputs.pr_head_sha, which doesn’t exist for this reusable workflow. That makes HEAD_SHA empty and causes orchestrator-reconcile-state.sh to fail the head SHA check, preventing state from being reconciled/upserted.

          # Reusable workflows retain the worker's event context. Bind state to that
          # context rather than trusting the agent-controlled callback payload.
          PR_NUMBER: ${{ github.event.inputs.pr_number }}
          HEAD_SHA: ${{ github.event.inputs.pr_head_sha }}
          OUTCOME: ${{ inputs.outcome }}