bmad-review-adversarial-general

An adversarial code review identifies weaknesses, omissions, and quality risks in a diff or branch by deliberately assuming issues are present. It outputs a Markdown list of at least ten findings to address before sharing with stakeholders.

Adversarial spec analysis critically examines specifications or stories to surface missing details and unsupported assumptions. You need it when you want a harsh quality pass on a product requirement or technical design before sharing it with stakeholders.

Yes, you can use this to find gaps in a technical document by applying a skeptical review style. It looks beyond obvious errors to identify missing details, suspicious omissions, and unsupported assumptions, returning a concise findings report.

The findings report uses a Markdown list format containing descriptions only. It generates at least ten findings focused on what should be fixed or improved in the reviewed content.

The review requires nonempty readable input but supports various content types. It automatically identifies the content type, applying to diffs, branches, uncommitted changes, specifications, stories, and documents.

A limitation of this skeptical quality assessment is that it requires nonempty readable text to function. It cannot evaluate empty inputs and focuses strictly on producing findings rather than rewriting or fixing the content itself.