What problem does it solve? Teams accumulate inconsistent conventions across a codebase, and ad-hoc style rules either go undocumented or get written as vague documents that drift out of date and get over-applied. This Skill produces durable, evidence-grounded coding standards that stay accurate as the code evolves and are surfaced to Claude Code only when relevant. ## Core Features & Use Cases - Three operating modes: create a new standard from scratch, convert an existing document (such as an ADR) into a standard, or update an existing standard while preserving its frontmatter. - Evidence-gathering agents: dispatches codebase-explorer agents to find real correct-usage and anti-pattern examples with durable, greppable anchors instead of bare line numbers. - YAGNI and adoption-bias audits: rejects standards better handled by linters, demands concrete evidence the pattern exists today, and structurally guards against over-application with decision trees and "when not to apply" sections. - Path-scoped integration: routes each standard into per-file-type index files under .claude/rules/coding-standards/ so Claude Code loads only a small index and opens full standards on demand. - Use Case: Ask it to standardize how Svelte stores shape state; it explores the codebase for existing store patterns, drafts the standard from real examples, runs adversarial reviews with junior-developer and information-architect agents, and wires it into the TypeScript/Svelte rule indexes. ## Quick Start Ask the assistant to create a coding standard for a topic such as "naming conventions for API route handlers" and let it gather codebase evidence before writing.
