roast

To critique architecture or documentation with evidence, run five personas against the target material to surface weaknesses, validate findings against repository evidence, and generate a prioritized improvement report with precise file:line citations.

Evidence-based stress testing for code and ideas is a devil's advocacy process that identifies issues by running multi-perspective critiques, tagging findings as validated claims, and verifying them against actual repository evidence.

No, you cannot use this for systematic code review or implementation changes. It provides read-only critique and stress testing of existing artifacts, so use dedicated code review tools for implementation changes or performance profiling.

To validate claims and get file:line references, apply critique personas to documentation, code, or architecture within the repository, tag each finding as [CLAIM-N], and validate it against the repository evidence for verification.

A read-only critique tool does not modify source files during analysis. It adheres to read-only constraints while evaluating targets, producing an actionable improvement plan and validated strengths without altering the repository.

The best way to poke holes in an architecture proposal is running five HackerNews commenter personas to surface blind spots, backing each critique with evidence-based claims validated against the repository for a prioritized report.