Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Scan for outdated or missing drivers - takes under a minute3Repair Windows errors before they cause bigger problemsA Claude Code custom skill packages a repeatable workflow in a directory containing a SKILL.md file. Put it in ~/.claude/skills/ for your own use or in .claude/skills/ inside a project when it should live with that repository. The file’s YAML frontmatter tells Claude what the skill does and when it is relevant; its Markdown body explains how to carry out the task.
Choose where the skill should live
Use the personal location for a workflow you want available across your own work, and the project location when the instructions belong with a particular repository. Anthropic documents these paths in its Agent Skills overview.
As an Amazon Associate I earn from qualifying purchases.
| Scope | Location | Practical use |
|---|---|---|
| Personal | ~/.claude/skills/ |
An individual workflow you want to reuse across projects. |
| Project | .claude/skills/ |
A workflow that should be stored with and shared through a project repository. |
The locations are documented by Anthropic; the distinction in use is a practical way to decide which scope fits your workflow.
Create the skill directory and its instruction file
Start with a directory named for the workflow and put SKILL.md at its root. For example, a project skill might be organized like this:
.claude/skills/review-change/
├── SKILL.md
├── references/ (optional)
├── scripts/ (optional)
└── assets/ (optional)The skill directory and its SKILL.md are fundamental. The other directories are optional categories for supporting material, not a required layout; include them only when the workflow needs them. Anthropic describes the folder design and supporting content in its Agent Skills overview and skills fundamentals guide.
Write useful frontmatter and focused instructions
Begin SKILL.md with YAML frontmatter. Anthropic’s general custom-skill guidance requires a name and description. The name identifies the skill; the description should say what it accomplishes and the kinds of requests for which it should be used, helping Claude decide when it is relevant. See Anthropic’s custom-skills guidance.
A minimal shape is:
---
name: review-change
description: Review a proposed code change for correctness and relevant project conventions. Use when asked to review a diff or change.
---
# Review a code change
1. Read the relevant project instructions and inspect the proposed change.
2. Check for correctness issues and violations of project conventions.
3. Report actionable findings with locations and explain any uncertainty.This is an illustrative starting point, not a required review procedure. Replace its name, description, and steps with the actual workflow. Keep the skill centered on one recurring task, use clear procedural instructions, and add examples when they clarify expected inputs or outputs.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Keep general limits distinct from Claude Code guarantees
Anthropic’s Help Center gives a 64-character maximum for the name and a 200-character maximum for the description in its general custom-skills guidance. Those figures should not be assumed to be Claude Code-specific limits unless the current Claude Code documentation confirms them. Claude Code’s supported fields and behavior can change, so check its current reference before relying on additional frontmatter fields.
Add references, scripts, or assets only when they help
Keep the essential workflow in SKILL.md. If a skill needs detailed material that is useful only for some tasks, put it in a reference file and point to it from the relevant instruction. Reusable files can go in an assets directory; scripts may support executable steps. This progressive-disclosure approach keeps the main instructions readable while making more detail available when needed. Anthropic describes the layers as metadata, the skill body, and linked files in its skills fundamentals guide.
Do not add executable code just because the folder structure allows it. A script adds behavior to review and maintain; use one only when it meaningfully supports the workflow.
Review the skill for safety before relying on it
Anthropic warns that skills can contain instructions and code that direct Claude to use tools in ways that go beyond what the description suggests. Treat a skill as executable guidance, not just documentation. Before using one, review its instructions and any scripts, compare their actual behavior with the declared purpose, and consider the effects of the tools or operations it may invoke. Anthropic’s warning appears in the Agent Skills overview.
Recommended Free Tools
- Read the complete
SKILL.mdand any linked files. - Inspect scripts before running or authorizing them, including their inputs and side effects.
- Check whether the skill’s requested actions fit its stated task and intended project scope.
- Use skills only when you trust their source and understand the actions they may guide.
Validate it with realistic requests
Anthropic’s cited guidance explains structure, metadata, supporting files, and safety, but does not establish a Claude Code-specific test protocol or official pass/fail criteria. A practical authoring check is to try representative requests, observe whether the intended skill is selected, inspect the resulting work, and verify any script side effects. Treat that as a useful validation routine, not an Anthropic-mandated test suite.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
