Claude Skills

A collection of Claude Code skills for enhanced AI-assisted development workflows.

Available Skills

Skill	Description
scope-check	Context window evaluation and chunking strategy tool. Prevents mid-implementation context collapse by validating document size before writing large design documents or implementation plans.
benji-taylor-design-eval	Structured design critique based on Benji Taylor's six principles (progressive disclosure, fluidity, simplicity/delight triad, craft, trust through clarity, accessibility). Evaluates any product or interface against these dimensions.

Installation

Via Claude Code Marketplace

/plugin marketplace add dvdarkin/claude-skills
/plugin install scope-check@dvdarkin

Restart Claude Code to pick up the new skill.

Manual Installation

Clone the repository and copy desired skills to your Claude skills directory:

git clone https://github.com/dvdarkin/claude-skills.git
cp -r claude-skills/skills/scope-check ~/.claude/skills/

Skills Overview

scope-check

Evaluates specification documents against Claude Code's context window constraints. Provides grounded estimations of whether a spec will fit within safe context bounds and recommends chunking strategies when it won't.

Features:

• Automatic activation before writing design documents or implementation plans
• Traffic light system (GREEN/YELLOW/ORANGE/RED) for context budget status
• Interactive chunking recommendations for large specs
• Research-backed expansion estimates based on language and TDD level

Quick Start:

# Basic check
/scope-check docs/feature-spec.md

# With explicit overrides
/scope-check docs/feature-spec.md --stack typescript --tdd comprehensive

Example Output: See scope-check-example.md for a real-world example analyzing a complex AI Virtual Car Showroom spec.

Research Foundation:

• QSM Database: Function Point to LOC ratios (2,192 projects)
• IFPUG ISO/IEC 20926:2009: Function Point counting standards
• Liu et al. (2024): "Lost in the Middle" - context position bias
• Chroma Research: "Context Rot" study across 18 LLMs
• RULER Benchmark: Long-context model evaluation

See scope-check documentation for full details.

benji-taylor-design-eval

Evaluates a product or interface against six design principles drawn from Benji Taylor's work (Family wallet, Base, X). Produces a structured critique covering progressive disclosure, fluidity, simplicity/fluidity/delight triad, craft, trust through clarity, and accessibility without depth sacrifice.

Best for: Consumer products and prosumer tools. Less directly applicable to pure developer tooling or enterprise software (flag and adapt rather than skip).

Quick Start:

Invoke the skill when reviewing any product design, UI screenshots, or mockups.
Ask: "evaluate this design" or "what would a great designer change here"

Contributing

Contributions welcome. If you have a skill that could benefit the Claude Code community, feel free to submit a PR.

Issues and PRs: https://github.com/dvdarkin/claude-skills

License

MIT License - see LICENSE for details.

Author

Denis Darkin (dvdarkin@gmail.com)