Software Development Skills

Use when designing AI features and the real opportunity may be perception support, not delegated conversational work.

This skill SHOULD be used for structured extraction or batch queries against JSON (use jq), YAML (use yq), or Markdown (use mq), and for advanced text search with ripgrep flags or pipe composition (use rg).

Audit and improve project memory files (CLAUDE.md, AGENTS.md, .claude.local.md) — assess against a quality rubric, then apply additions and removals. This skill SHOULD be used when the user asks to audit, improve, edit, fix, tighten, rewrite, or update a memory file, or to check whether one is too long, stale, or bloated.

Clean-room reimplementation of a repository, library, API, protocol, or other software artifact. This skill SHOULD be used when the user wants to create a functionally equivalent implementation of an existing artifact without copying its internal design or expression. NOT for refactoring existing code.

Lift-and-shift port of a repository, library, API, protocol, or other software artifact to a different language, runtime, or platform, with the original source available as a reference. This skill SHOULD be used when the user says "port from X to Y", "lift and shift", "rewrite in Go", "rewrite in Rust", "translate this codebase", "convert to TypeScript", "language migration", "rewrite in another language", "carry over the algorithms", "borrow techniques from", "dirty room", "dirty-room design", "sc-dirty-room", or wants to reimplement an existing artifact while studying its internals to preserve algorithms, data structures, or idioms worth keeping. NOT for legally-constrained reimplementation — use /sc-clean-room when copyright isolation matters.

Generate and edit images using the Gemini API (Nano Banana). This skill SHOULD be used when creating images from text prompts, editing existing images, applying style transfers, generating logos with text, creating stickers, product mockups, or any image generation/manipulation task. Supports text-to-image, image editing, multi-turn refinement, and composition from multiple reference images.

Use when multiple plausible approaches must be generated, tested, and compared against constraints before the user chooses.

Generate mermaid diagrams — flowcharts for architecture, sequence diagrams for interactions. This skill SHOULD be used when the user asks to "create a diagram", "draw an architecture diagram", "sequence diagram", "flowchart", "visualize the system", or "generate mermaid".

Use when the user asks to "review code", "find dead code", "check for duplication", "simplify the codebase", "find refactoring opportunities", "do code cleanup", "check naming consistency", "analyze test organization", "run codebase health check", "review my PR", "refactor this code", "extract method", "rename variable", "consolidate duplicates", "adversarial review", "red team review", "find ways to break this", "multi-model review", "get multiple AI opinions on this code", "hunt bugs", "find bugs", "bug hunt", or "adversarial bug hunt". Routes to specialized analysis agents, refactoring workflow, multi-model adversarial review, or adversarial bug hunt based on the type of request.

Use when the user is clarifying beliefs, assumptions, goals, or framing before committing to decision or plan.

Reframe code design through functional programming principles for agent-assisted development. This skill SHOULD be used when the user says "think functional", "think FP", "make this pure", "separate effects", "where should this side effect go", "this function does too much", "how should I structure this for agents", "make this easier to review", "reduce context needed", or when planning module structure, store design, or code that agents will write and humans will review. Applies FP discipline within any language to maximize agent effectiveness and human reviewability.