How to use the Embodiment Description Writer agent skill

What is an agent skill, and what does this one do?

An agent skill is a folder containing a SKILL.md file that gives an agent like Claude Code a set of instructions for a specific task. When the task comes up, the agent follows those instructions instead of improvising.

This skill, named embodiment-description, tells the agent to write one to three detailed embodiments that support your drafted claims. An embodiment describes how to make and use the invention. For each embodiment the agent plans the structure, describes each component or method step with reference numerals tied to figures, adds variations and alternatives, and checks that every claim element is supported by at least one embodiment.

• Plans embodiments that cover the claims, including at least one best-mode embodiment.
• Describes components or method steps with numerals such as processor (102) and FIG. 1.
• Adds variations such as alternative materials, parameters or configurations.
• Checks each claim element against the embodiments and adds missing description.
• Writes output to patent/specification/detailed_description.md.

Install the embodiment-description skill

Download the ZIP for skills/embodiment-description from this page. Unzip it so that the folder containing SKILL.md ends up in one of the locations below. Choose one of the two options.

For a personal skill available in all your projects, unzip it into ~/.claude/skills/. For a skill that only applies to one project, unzip it into the .claude/skills/ folder inside that project.

If you use the Claude apps rather than Claude Code, you can upload the skill under Settings → Capabilities → Skills instead.

• Personal install: ~/.claude/skills/embodiment-description/SKILL.md
• Project install: .claude/skills/embodiment-description/SKILL.md inside your project folder
• Claude apps: Settings → Capabilities → Skills, then upload the skill

Prepare your input files

The skill reads your invention material from fixed paths. Create a patent/ folder in your project and place your files there before you ask for an embodiment. The skill's output quality depends heavily on these inputs.

The skill expects patent/INVENTION_DISCLOSURE.md, which breaks the invention into core, supporting and optional features, and patent/CLAIMS.md, which holds the drafted claims the embodiments must support. If you have figures, the skill also accepts a patent/figures/numeral_index.md file, which comes from the figure-description skill. Without the disclosure and claims files, the skill's results will be thin or guessed.

• patent/INVENTION_DISCLOSURE.md (recommended)
• patent/CLAIMS.md (recommended)
• patent/figures/numeral_index.md (optional, from the figure-description skill)
• User-supplied figures in any directory

Trigger the skill with natural requests

You do not need a special command. Ask in plain language, and the agent will load the skill when your request matches its description. The skill responds to phrases such as write embodiment, detailed description, 实施例描述 and 撰写实施例, and to any request to describe how to practice an invention.

The skill also accepts an argument hint of claims-path-or-embodiment-details, so you can point it at a claims file or describe the embodiment you want.

• write embodiment for the claims in patent/CLAIMS.md
• Draft the detailed description for my invention, the sensor uses a resonant frequency shift
• 撰写实施例，基于 patent/CLAIMS.md
• Describe how to practice the invention in claims 1 to 5

Review what the skill produces

The skill writes each embodiment as a self-contained section and cross-references other embodiments when describing alternatives. Numerals follow a figure-series scheme, with 100-series for FIG. 1 and 200-series for FIG. 2. The first mention of a component carries its numeral in parentheses, and later mentions do not.

The skill is explicit about what an embodiment should not contain. It describes the invention and does not report test results, accuracy figures or comparisons with prior art. Check the draft against that rule and against your claims before you rely on it.

• First mention: the processor (102). Later mentions: the processor 102.
• Software embodiments use pseudocode or algorithm steps, not actual source code.
• Experimental results, accuracy percentages and performance comparisons are excluded by design.
• Every claim element should appear in the claim-support table the skill builds.

Safety notes: scripts and network

According to its listing, this skill runs no scripts and uses no network access. Its only file is SKILL.md. Its tool permissions cover Bash, Read, Write, Edit, Grep and Glob, so the agent can read and write files in your project while it works. Review any file it writes before you rely on it.

The output is a drafting aid. The skill itself warns that enablement and best-mode requirements depend on jurisdiction and facts it cannot verify, so a registered patent professional should review the final text.

• Runs scripts: no
• Uses network: no
• Reads and writes project files, so keep backups of your patent folder.
• Do not paste confidential material into an environment you do not trust.