Cartograf666
UserPortable PRD-first, architecture- and design-system-aware Agent Skills for verified multi-model development across Claude, Gemini, and OpenAI Codex.
Categories
Indexed Skills (5)
architecture-governance
Bootstrap, document, enforce, and review a project's architecture through layer responsibilities, dependency direction, architecture manifests, legacy baselines, Change Envelopes, cross-layer approval policy, and deterministic architecture checks. Use when starting a project, auditing an existing codebase, planning cross-layer work, preventing architecture erosion, reviewing structural changes, or deciding whether implementation may expand into another layer.
design-system-governance
Discover, select, document, extend, and enforce a product design system through design tokens, component contracts, responsive behavior, interaction states, content rules, accessibility, visual evidence, and deterministic UI quality gates. Use when starting a UI product, preparing a frontend feature, choosing a component system or visual direction, auditing an existing interface, preventing one-off styling, or reviewing whether implementation follows the approved design language.
grill-requirements
Resolve product, UX, technical, architecture, risk, and scope decisions through a disciplined one-question-at-a-time interview before implementation. Use when the user asks to grill, challenge, stress-test, clarify, interrogate, or pressure-test an idea, PRD, feature, project plan, architecture proposal, or ambiguous development request.
prepare-development-cycle
Convert a new product idea, feature request, bug, refactor, or architecture change into an evidence-backed and user-approved PRD, Technical Brief, RunSpec, and architecture-aware Change Envelope before implementation. Use when starting a new project, preparing a development task, writing or refining a PRD, clarifying requirements, scoping work, or creating a handoff for an automated coding cycle.
run-verified-development-loop
Execute an approved PRD and development contract through bounded multi-model implementation cycles with architecture-aware work packages, isolated worktrees, separate coding and testing roles, deterministic architecture and quality gates, repair limits, and independent review. Use when the user asks to implement approved work automatically with an orchestrator such as Fable, Opus, or OpenAI Codex; workers such as Sonnet, Gemini, or Codex; testers such as Haiku; or explicit requirements that the result be structurally clean, tested, and working.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.