Patent Concept Extraction

What it does

This skill extracts the core technical concepts and inventive approaches from patent claim text. Rather than restating claims, it identifies the underlying technical contributions that make an invention non-obvious, helping patent professionals, examiners, and researchers quickly grasp the novel concepts. The skill is built to work by loading a prepared prompt template, soliciting required inputs (patent claims), filling placeholders, and executing the prompt to produce structured concept extractions.

When to use it

Use this skill when reviewing patent claims to identify inventive features, preparing prosecution notes, prior-art searches, or summarizing technical approaches for litigation or licensing. It is helpful when you need concise, technically-focused concept summaries rather than verbose claim restatements.

What's included

• Scripts: reads a prompt template located in the repo and uses it to run extraction (the workflow expects the environment variable CLAUDE_SKILL_DIR to locate prompt files).
• References: none bundled, but the repo structure contains a prompt file referenced by the skill.
• Instructions: stepwise process — load prompt template, extract the prompt section, ask the user for patent claims if not provided, substitute placeholders, execute the prompt, and present extracted concepts.

Compatible agents

Designed for instruction-following language agents (Claude, Copilot-like assistants) that can run prompt-based workflows and accept user-provided claim text.