fix: add copy-paste command for adversarial review in fresh context#1926
fix: add copy-paste command for adversarial review in fresh context#1926sunilp wants to merge 5 commits intobmad-code-org:mainfrom
Conversation
📝 WalkthroughWalkthroughUpdated step-04-review.md in the bmad-quick-flow to provide a copy-paste command for running Adversarial Review in a fresh context. The change reorders the final menu to emphasize adversarial review first, introduces an explicit Changes
Estimated code review effort🎯 2 (Simple) | ⏱️ ~8 minutes Possibly related PRs
Suggested reviewers
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
📝 Coding Plan for PR 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. Comment |
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md (1)
131-135:⚠️ Potential issue | 🟠 MajorBring
[B]in line with the fresh-context recommendation.The new copy says implementation should run in a fresh context, but selecting
[B]still starts development from the current session. That leaves the menu behavior contradicting the recommendation you just added. Either make[B]print the copy-pastequick-dev {finalFile}command and halt, or weaken the surrounding claim.Also applies to: 145-145
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md` around lines 131 - 135, The menu option labelled "[B]" currently starts development in the current session which contradicts the new guidance to run implementation in a fresh context; update the "[B]" behavior to instead print the exact copy-paste command "quick-dev {finalFile}" and exit (no session start), or if you prefer to keep the interactive path, change the surrounding copy that mandates a fresh context so it no longer states implementation must be run in a fresh context; ensure the change references the "[B]" menu option and the "quick-dev {finalFile}" command so reviewers can locate and verify the fix.
🧹 Nitpick comments (5)
src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md (5)
125-129: Pass the workflow’s acceptance lens into the adversarial review.
bmad-review-adversarial-generalaccepts optionalalso_consider, but the pasted command only sends{finalFile}. Passing the READY FOR DEVELOPMENT standard here would make the findings better aligned with the actual bar this step is enforcing.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md` around lines 125 - 129, The adversarial review command bmad-review-adversarial-general is being invoked with only {finalFile}; include the workflow acceptance lens by adding the optional also_consider argument (e.g., also_consider="READY FOR DEVELOPMENT") so the reviewer uses the correct acceptance standard; update the example invocation to pass also_consider alongside {finalFile} to align findings with the step’s required bar.
125-129: Tell the user to open a new session explicitly.“Fresh context” is BMAD jargon, not an action. Add a direct instruction like “open a new chat/session and run:” so users do not paste this back into the same conversation and defeat the purpose.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md` around lines 125 - 129, Update the guidance in step-04-review.md to explicitly instruct the user to open a new chat/session before running the adversarial review: replace or augment the phrase "fresh context" with a clear action such as "Open a new chat/session and run:" followed by the existing command example (bmad-review-adversarial-general {finalFile}) so users do not run it in the same conversation; ensure the text references the command symbol bmad-review-adversarial-general and the context instruction is adjacent to the example so it is unambiguous.
125-129: The external review command has no return path.Users are told to run adversarial review externally, then immediately shown the dev command, but nothing here says what to do with the findings. Add one sentence telling them to bring the findings back to this workflow for triage/apply/rerun before moving to
quick-dev.Also applies to: 131-135
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md` around lines 125 - 129, After running the external adversarial review command bmad-review-adversarial-general {finalFile}, add a single clear sentence instructing the user to bring the review findings back into this workflow for triage and to apply or rerun fixes (e.g., “Bring the reviewer’s findings back into this workflow for triage/apply/rerun before proceeding to quick-dev.”); make the same edit to the second occurrence of this external-review block so both places instruct users to triage/apply/rerun findings locally before moving on to quick-dev.
131-135: Say when to rerun adversarial review after edits.The new flow mentions adversarial review, then jumps to implementation, but this menu still allows further edits via Advanced Elicitation and Party Mode. A short note to rerun adversarial review after material spec changes would close that loop instead of implying the review is one-and-done.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md` around lines 131 - 135, Add a short clarifying sentence to the "Once you are fully satisfied..." paragraph in step-04-review.md that instructs users to rerun the Adversarial Review after any material edits made via Advanced Elicitation or Party Mode before running implementation; reference the existing "Adversarial Review", "Advanced Elicitation", and "Party Mode" terms and the "quick-dev {finalFile}" command so the note appears right before the code block to close the loop and avoid implying the review is one-and-done.
131-135: Carry the new adversarial-review guidance into the exit flow.If the user chooses
[D]right away, the exit message still only advertisesquick-dev {finalFile}. That drops the exact command this PR is trying to surface and makes the new UX easy to miss. Mirror the adversarial-review command in Section 5 as well.Also applies to: 173-185
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md` around lines 131 - 135, Update the exit flow so the final exit message advertises the adversarial-review command as well as quick-dev: when the user picks option "[D]" (the immediate done/exit path) the message should include both "adversarial-review {finalFile}" and "quick-dev {finalFile}" (mirror the adversarial-review guidance you added earlier), and make the same change in Section 5's exit text so both sections consistently surface the adversarial-review command alongside quick-dev.
🤖 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/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md`:
- Around line 125-129: The wording for "Adversarial Review" overstates
isolation; update the paragraph that introduces the
bmad-review-adversarial-general {finalFile} command to either (A) explicitly
state that the reviewer should be limited to only reviewing {finalFile} (i.e.,
instruct reviewers to avoid project context) or (B) soften the claim to note
that the reviewer should "only be provided the spec ({finalFile}) when possible"
and acknowledge that project read access may still exist; apply the same wording
change to the analogous sentence elsewhere in the file where the same claim
appears.
---
Outside diff comments:
In `@src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md`:
- Around line 131-135: The menu option labelled "[B]" currently starts
development in the current session which contradicts the new guidance to run
implementation in a fresh context; update the "[B]" behavior to instead print
the exact copy-paste command "quick-dev {finalFile}" and exit (no session
start), or if you prefer to keep the interactive path, change the surrounding
copy that mandates a fresh context so it no longer states implementation must be
run in a fresh context; ensure the change references the "[B]" menu option and
the "quick-dev {finalFile}" command so reviewers can locate and verify the fix.
---
Nitpick comments:
In `@src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md`:
- Around line 125-129: The adversarial review command
bmad-review-adversarial-general is being invoked with only {finalFile}; include
the workflow acceptance lens by adding the optional also_consider argument
(e.g., also_consider="READY FOR DEVELOPMENT") so the reviewer uses the correct
acceptance standard; update the example invocation to pass also_consider
alongside {finalFile} to align findings with the step’s required bar.
- Around line 125-129: Update the guidance in step-04-review.md to explicitly
instruct the user to open a new chat/session before running the adversarial
review: replace or augment the phrase "fresh context" with a clear action such
as "Open a new chat/session and run:" followed by the existing command example
(bmad-review-adversarial-general {finalFile}) so users do not run it in the same
conversation; ensure the text references the command symbol
bmad-review-adversarial-general and the context instruction is adjacent to the
example so it is unambiguous.
- Around line 125-129: After running the external adversarial review command
bmad-review-adversarial-general {finalFile}, add a single clear sentence
instructing the user to bring the review findings back into this workflow for
triage and to apply or rerun fixes (e.g., “Bring the reviewer’s findings back
into this workflow for triage/apply/rerun before proceeding to quick-dev.”);
make the same edit to the second occurrence of this external-review block so
both places instruct users to triage/apply/rerun findings locally before moving
on to quick-dev.
- Around line 131-135: Add a short clarifying sentence to the "Once you are
fully satisfied..." paragraph in step-04-review.md that instructs users to rerun
the Adversarial Review after any material edits made via Advanced Elicitation or
Party Mode before running implementation; reference the existing "Adversarial
Review", "Advanced Elicitation", and "Party Mode" terms and the "quick-dev
{finalFile}" command so the note appears right before the code block to close
the loop and avoid implying the review is one-and-done.
- Around line 131-135: Update the exit flow so the final exit message advertises
the adversarial-review command as well as quick-dev: when the user picks option
"[D]" (the immediate done/exit path) the message should include both
"adversarial-review {finalFile}" and "quick-dev {finalFile}" (mirror the
adversarial-review guidance you added earlier), and make the same change in
Section 5's exit text so both sections consistently surface the
adversarial-review command alongside quick-dev.
🪄 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: 3bed2c8f-31e2-4356-8f89-2c3e8d2fc55a
📒 Files selected for processing (1)
src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md
| For best results, run **Adversarial Review** in a fresh context to ensure information asymmetry (the reviewer has no knowledge of your design decisions, only the spec): | ||
|
|
||
| Copy this prompt to start dev: | ||
| \`\`\` | ||
| bmad-review-adversarial-general {finalFile} | ||
| \`\`\` |
There was a problem hiding this comment.
The isolation claim is stronger than the workflow actually guarantees.
This says the reviewer has “only the spec,” but the [R] process still allows a reviewer with project read access. Either constrain the recommendation to review only {finalFile} or soften the wording so the prompt is not promising stricter isolation than the workflow provides.
Also applies to: 159-160
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md` around
lines 125 - 129, The wording for "Adversarial Review" overstates isolation;
update the paragraph that introduces the bmad-review-adversarial-general
{finalFile} command to either (A) explicitly state that the reviewer should be
limited to only reviewing {finalFile} (i.e., instruct reviewers to avoid project
context) or (B) soften the claim to note that the reviewer should "only be
provided the spec ({finalFile}) when possible" and acknowledge that project read
access may still exist; apply the same wording change to the analogous sentence
elsewhere in the file where the same claim appears.
- [B] now prints copy-paste command instead of starting dev in current session - Soften isolation claim in [R] process to acknowledge project read access - Add guidance to rerun adversarial review after material spec changes - Add adversarial-review command to exit flow (Section 5)
What
Adds a copy-paste command for running adversarial review in a fresh context, matching the existing UX pattern for the quick-dev command.
Why
In quick-spec step-04-review.md, the final menu provides a ready-to-copy command for starting development in a fresh context, but no equivalent for adversarial review. The workflow recognizes fresh context matters for dev but does not apply the same pattern for adversarial review, which also benefits from information asymmetry.
Fixes #1659
How
Testing