fix(code-review): restore actionable review output with interactive choices

[alexeyv]

Summary

• Restores the interactive post-review action menu removed in PR feat(skills): rewrite code-review with sharded step-file architecture #2007 (March 15 rewrite)
• Deferred findings: auto-written to deferred-work.md and checked off in story file — no HITL turn needed
• Intent gap / bad spec findings: conversation with options ordered cheapest-to-most-expensive: downgrade to patch, patch the spec, reset to ready-for-dev, dismiss
• Patch findings: three choices restored — fix automatically, create action items ([AI-Review][Severity]), or show details

Motivation

The rewrite made the review step presentation-only, removing the ability to act on findings. This forced expensive upstream re-runs (create story = ~15 HITL turns, dev story = full cycle) even when a simple code patch would suffice. The new "downgrade to patch" option lets users reclassify spec findings into the patch bucket to avoid unnecessary re-runs.

Test plan

• Run code review with a spec file that produces patch findings — verify 3-option menu appears
• Run code review that produces defer findings — verify auto-write to deferred-work.md and story file
• Run code review with intent_gap findings — verify conversation with 4 options (downgrade, patch spec, reset, dismiss)
• Choose "downgrade to patch" for an intent_gap — verify it flows into patch menu
• Choose "create action items" — verify [AI-Review][Severity] format in story file

🤖 Generated with Claude Code

[seynurmammad0v]

Thanks, I caught this as well.

[github-actions]

@coderabbitai review

[augmentcode]

🤖 Augment PR Summary

Summary: Restores actionable behavior to the code review workflow’s final step by reintroducing interactive “what next?” decisions after findings are presented.

Changes:

• Renames Step 4 to “Present and Act” and adds clearer handling rules by finding type
• Automatically writes defer findings to deferred-work.md and optionally marks them complete in the story file
• Adds an intent-gap / bad-spec resolution conversation with options ordered from lowest to highest cost (including “downgrade to patch”)
• Restores a 3-option patch findings menu (auto-fix, create action items, or walkthrough details)

Technical Notes: Introduces a deferred_work_file frontmatter variable pointing at {implementation_artifacts}/deferred-work.md to standardize deferred output.

_{🤖 Was this summary useful? React with 👍 or 👎}

[augmentcode]

Review completed. 4 suggestions posted.

Comment augment review to trigger a new review at any time.

[coderabbitai]

📝 Walkthrough

Walkthrough

Updated Step 4 workflow instructions for "Present and Act" to automatically append defer findings to a configurable file, introduce a clean review shortcut when no findings remain, and restructure the presentation and handling of findings with interactive control flow for patches and intent gaps.

Changes

Cohort / File(s)

Summary

Step 4 Workflow Instructions src/bmm-skills/4-implementation/bmad-code-review/steps/step-04-present.md

Automated handling of defer findings (appended to {deferred_work_file} with standardized heading), new clean review shortcut to exit when no findings remain, restructured finding presentation by category with interactive control flow, intent-gap/bad-spec findings require conversation with explicit resolution options, patch findings require user choice via menu, and updated summary line wording to reflect auto-managed defer findings.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~8 minutes

Possibly related PRs

• fix(skills): address code-review findings from PR #2007 #2013: Modifies the same workflow step file and addresses how patch/defer findings are presented and handled in the code review process.

Suggested reviewers

• bmadcode

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately describes the main change: restoring interactive post-review action capabilities that were previously removed.
Description check	✅ Passed	The description is directly related to the changeset, providing clear motivation, summary of changes, and a comprehensive test plan for the restored functionality.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix-code-review-actions

📝 Coding Plan

• Generate coding plan for human review comments

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🧹 Nitpick comments (1)

src/bmm-skills/4-implementation/bmad-code-review/steps/step-04-present.md (1)

16-18: Consider acknowledging failed layers in the clean review shortcut.

Step-03 triage (lines 41-46 in step-03-triage.md) warns the user when zero findings remain but {failed_layers} is non-empty, indicating the review may be incomplete. This shortcut doesn't reference that warning. Consider adding a note that the clean review shortcut assumes step-03 already handled the failed layers warning.

📝 Suggested clarification

 ### 1. Clean review shortcut
 
-If zero findings remain after triage (all rejected or none raised): state that and end the workflow.
+If zero findings remain after triage (all rejected or none raised): state that and end the workflow. (Note: step 3 will have already warned if any review layers failed.)

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@src/bmm-skills/4-implementation/bmad-code-review/steps/step-04-present.md`
around lines 16 - 18, Update the "Clean review shortcut" text in
step-04-present.md to acknowledge failed layers: either add a short note that
this shortcut assumes step-03 triage has already handled any {failed_layers}
warnings, or explicitly instruct the reviewer to verify {failed_layers} is empty
before ending the workflow; reference the Step-03 triage behavior mentioned in
step-03-triage.md (the warning about non-empty {failed_layers}) so readers know
where to check.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@src/bmm-skills/4-implementation/bmad-code-review/steps/step-04-present.md`:
- Around line 57-68: The Option 2 flow assumes a non-empty spec_file and will
fail when review_mode == "no-spec"; update the Option 2 handling in the step
that presents "How would you like to handle the {Z} patch findings?" to first
check that spec_file is present (non-empty) before attempting to add a "Review
Follow-ups (AI)" subsection to {spec_file}; if spec_file is empty (review_mode
== "no-spec") present a clear alternative (e.g., ask the user to choose another
option, create a downloadable patch/summary, or store follow-ups in a temporary
in-memory or output-only location) and do not attempt to write to {spec_file}.
Ensure the guard references the existing variables review_mode and spec_file and
is applied in the code path that implements Option 2.

---

Nitpick comments:
In `@src/bmm-skills/4-implementation/bmad-code-review/steps/step-04-present.md`:
- Around line 16-18: Update the "Clean review shortcut" text in
step-04-present.md to acknowledge failed layers: either add a short note that
this shortcut assumes step-03 triage has already handled any {failed_layers}
warnings, or explicitly instruct the reviewer to verify {failed_layers} is empty
before ending the workflow; reference the Step-03 triage behavior mentioned in
step-03-triage.md (the warning about non-empty {failed_layers}) so readers know
where to check.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: bfa38c4a-fef8-4515-89c3-9a3f9c864695

📥 Commits

Reviewing files that changed from the base of the PR and between 4cb58ba and 0277381.

📒 Files selected for processing (1)

• src/bmm-skills/4-implementation/bmad-code-review/steps/step-04-present.md