sharingan

What problem does it solve? Converting knowledge from external sources—GitHub plugin repos, technical articles, papers, or local code—into properly formatted sd0x-dev-flow skill definitions is manual and error-prone. This Skill automates the scan, analysis, generation, and validation pipeline so replicated skills conform to the target format with verified frontmatter, routing signatures, and dependency checks. ## Core Features & Use Cases - Multi-source ingestion: Accepts GitHub URLs (deterministic regex fast-path), non-GitHub URLs and concepts (LLM classifier with confidence gate), or local code paths, all normalized into a SourceBundle intermediate format. - Dependency-aware generation: Builds a skill dependency graph using Tarjan's SCC cycle detection and Kahn's topological sort to produce leaf-first batch generation plans. - 3-layer validation: Validates generated skills via frontmatter schema checks (L1), skill-lint format rules (L2), and LLM semantic consistency checks (L3) that flag hallucinated tools or missing references. - Security envelope: Enforces HTTPS-only fetching, denies private/reserved addresses, caps payloads at 500KB, sanitizes external content, and never executes fetched code. - Use Case: Point it at a community plugin repo like anthropics/skills to produce an analysis report with a dependency graph and untranslatable-element flags, then run generate mode to write validated SKILL.md files into your project. ## Quick Start Analyze the skills in the GitHub repo https://github.com/anthropics/skills and generate equivalent sd0x-dev-flow skill definitions in my skills directory.