القائمة
User-invoked only; use advisor solely when the user explicitly requests advisor.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Use when the user wants to design, redesign, shape, critique, audit, polish, clarify, distill, harden, optimize, adapt, animate, colorize, extract, or otherwise improve a frontend interface. Covers websites, landing pages, dashboards, product UI, app shells, components, forms, settings, onboarding, and empty states. Handles UX review, visual hierarchy, information architecture, cognitive load, accessibility, performance, responsive behavior, theming, anti-patterns, typography, fonts, spacing, layout, alignment, color, motion, micro-interactions, UX copy, error states, edge cases, i18n, and reusable design systems or tokens. Also use for bland designs that need to become bolder or more delightful, loud designs that should become quieter, live browser iteration on UI elements, or ambitious visual effects that should feel technically extraordinary. Not for backend-only or non-UI tasks.
SDD-mode PR-boundary review policy for Pi. The root launches visible reviewer lanes, publishes triage, ends the turn for acknowledgement, then fixes in a separate follow-up. Review is independent of CI.
Use when the user wants to design, redesign, shape, critique, audit, polish, clarify, distill, harden, optimize, adapt, animate, colorize, extract, or otherwise improve a frontend interface. Covers websites, landing pages, dashboards, product UI, app shells, components, forms, settings, onboarding, and empty states. Handles UX review, visual hierarchy, information architecture, cognitive load, accessibility, performance, responsive behavior, theming, anti-patterns, typography, fonts, spacing, layout, alignment, color, motion, micro-interactions, UX copy, error states, edge cases, i18n, and reusable design systems or tokens. Also use for bland designs that need to become bolder or more delightful, loud designs that should become quieter, live browser iteration on UI elements, or ambitious visual effects that should feel technically extraordinary. Not for backend-only or non-UI tasks.
Create a pull request with review, REQ backlinks, push, and CI handoff.
Verify a deployed UI in a browser for interaction, mobile layout, and visuals.
Workflow for /sdd clean — rescuing a rotted spec. Mode-aware behaviors (interactive/auto/unleashed), safety nets, what gets cleaned (incl. per-AC @impl AND @test anchor backfill at parity, gated by enforce_tdd), JUDGMENT auto-resolution rules. Invoked when /sdd clean runs. Requires the spec-driven-development skill for REQ format and Status semantics, and the spec-enforce skill family for the detection mechanics.
| name | advisor |
| description | User-invoked only; use advisor solely when the user explicitly requests advisor. |
The advisor tool (provided by the @juicesharp/rpiv-advisor extension) hands the entire current conversation branch to a stronger reviewer model and returns guidance.
Only the user may invoke advisor. Do not call the advisor tool, run /advisor, or suggest /advisor unless the user's current message explicitly asks for advisor, for example:
/advisorThe assistant must not invoke advisor proactively for:
If advisor might help but the user did not explicitly ask for it, continue without advisor or ask a normal clarification question. Do not mention advisor as a required step, and do not tell the user to run /advisor proactively.
Advisor authenticates through Pi's model registry and uses the model selected by the user through the user-only /advisor command. The selection persists across sessions at ~/.config/rpiv-advisor/advisor.json. Only the user should open /advisor; the assistant must not run, simulate, or recommend that command unless asked.
When the user's current message explicitly requests advisor, call advisor with no parameters. Relay the returned guidance and make clear what, if anything, it changes.
If the advisor tool is unavailable or returns an error, report that directly. Do not retry repeatedly and do not substitute another external LLM unless the user explicitly asks for that.