{"id":"stevembarclay-pencilplaybook","name":"pencilplaybook","af_score":49.0,"security_score":31.0,"reliability_score":21.2,"what_it_does":"PencilPlaybook is a Claude Code “skill” intended to guide Claude in producing UI/design outputs inside Pencil.dev by embedding perceptual design defaults (token lookup tables), structured workflows for working with .pen canvases, and presets/scaffolds for common screen archetypes. It also references a setup wizard that configures design-system presets and claims to provide a complete parameter reference for 12 Pencil MCP tools.","best_when":"You are actively using Pencil.dev with Claude Code and want repeatable, token-driven design guardrails and scaffold workflows rather than ad-hoc prompts.","avoid_when":"You need a documented, independently verifiable REST/SDK interface, or you cannot tolerate Claude/skill-driven modifications to your .pen project.","last_evaluated":"2026-04-04T19:33:04.761864+00:00","has_mcp":true,"has_api":false,"auth_methods":["Likely local skill-based configuration for Claude Code + Pencil MCP tools (exact auth method not described in provided README)"],"has_free_tier":false,"known_gotchas":["README notes MCP can disappear or re-inject and advises running setup once then pinning the skill; also suggests avoiding opening the Pencil desktop app while Claude Code is active because it can override MCP configs.","README suggests prefixing prompts with 'Using PencilPlaybook,' to ensure the skill’s workflows are applied.","Token spikes on large canvases; recommends canvas archaeology + bulk inspection before big refactors."],"error_quality":0.0}