review

# Review Run a review of the current context (code, PR, or docs). Your reply must be a **plan of suggested changes**: concise, actionable, and structured-not only prose. ## Invocation Text after the slash command is additional scope/focus - narrow the review accordingly. If none given, use the default scope described below. ## Best practices alignment This command is designed to align with industry best practices for (AI-assisted) code and technical reviews: - **Explicit scope and dimensions** - Clear categories (architecture, scalability, performance, security, maintainability, CI) so feedback is scoped and not generic. - **Context and conventions** - Reference to AGENTS.md and key artifacts ensures context-aware, project-aligned suggestions. - **Actionable, explainable feedback** - Every item requires **what**, **where**, and **why**; suggestions must be implementable. - **Structured output** - Critical / Improvements / Optional with headings and bullets for scannable plans. - **Conduct and scope** - Be constructive and specific. Stay in scope; call out out-of-scope concerns as separate follow-up items. For minor polish, prefix with **Nit:** so the author knows it's non-blocking. - **Trade-offs** - When a suggestion involves trade-offs (e.g. performance vs readability), mention them briefly so the author can decide. ## Deep technical review When asked for a full or deep codebase review, conduct a deep, end-to-end technical review with the objective of bringing the ...