Guide · Agent skill
How to use the Andru.ia Skill-Smith agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
Andru.ia Skill-Smith is an agent skill that helps an AI agent create or improve another reusable skill. It is aimed at maintainers, developers writing their own SKILL.md skills, and contributors preparing a pull request. After this guide you will know what the skill does, how to install it, how to ask for it in plain language, and what checks to run before proposing changes.
Andru.ia Skill-Smithsickn33/agentic-awesome-skills · MIT Download the skill Quick start
- 1Download and unzip. Download the ZIP from this page and unzip it so that SKILL.md sits in a skill folder.
- 2Place the skill. Move the folder to ~/.claude/skills/ or to a project's .claude/skills/, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Prepare the repository. Clone the target repository locally and confirm that docs/contributors/skill-template.md is available.
- 4Describe the skill. Ask the agent in plain language for a skill, giving its name, expert role and expected outputs.
- 5Review the draft. Check the generated SKILL.md for examples, limits and the expected results before accepting it.
- 6Run the checks. In the Andru.ia repository, run npm run chain, npm run validate:references, npm run security:docs and npm run test, and fix any failures at the source.
- 7Propose the pull request. Once the checks pass, have the agent prepare the pull request for the source files.
What an agent skill is and what Skill-Smith does
An agent skill is a folder containing a SKILL.md file with instructions that an AI agent can load when a task matches. Skill-Smith is a skill whose job is to write those instructions for a new skill.
When you ask the agent to create or improve a reusable skill, Skill-Smith makes it follow a structured process. It first pins down the skill's name, expert role and expected outputs, then drafts a SKILL.md with examples and limits, and finally runs the repository's own checks before proposing a pull request.
The skill is written in Spanish by default but follows the language you request. It only drafts and prepares the work; it does not install or publish anything on its own.
Install the skill
Download the ZIP from the skill's page on this directory and unzip it. The skill folder contains the SKILL.md file.
For a personal install, place the unzipped folder in ~/.claude/skills/. For a single project, place it in the project's .claude/skills/ folder.
In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal: ~/.claude/skills/
- Project: .claude/skills/ inside your project
- Claude apps: Settings → Capabilities → Skills (upload)
Check what you need before you start
Skill-Smith is designed around the Andru.ia repository. Its instructions reference the repository's own checks, so you need a local clone of the target repository and Git.
For the Andru.ia checks you also need Node.js and npm, and access to docs/contributors/skill-template.md in the repository. The skill does not run scripts itself and does not use the network.
- Git and a local clone of the target repository
- Node.js and npm (for the Andru.ia checks)
- Access to docs/contributors/skill-template.md in the repository
Trigger the skill with a natural request
You do not need a special command. Describe the skill you want: its purpose, the inputs it takes, the results it should produce and how success is judged. The skill asks for the three basic pillars if you leave them out: a technical name, an expert role, and key outputs.
Use the skill for authoring only. It is not meant for performing tasks such as data checks or document work.
- Name: a technical identifier such as csv-quality-check (in Andru.ia, lowercase words joined by hyphens)
- Role: who the skill's expert is, for example an expert in security auditing
- Outputs: which files or actions the skill should produce
Example: a CSV quality-check skill
Suppose you ask: Crea una skill csv-quality-check que informe valores ausentes y duplicados sin modificar el CSV. The skill's illustrative design takes an authorised path, a delimiter and key columns as inputs, and reports the row count, missing values per column and repeated keys.
As an illustration, the rows id,name / 1,Ana / 1, produce one duplicate key and one missing name, and the CSV stays unchanged. An ambiguous delimiter is recorded as an unresolved input, not as a clean report.
Crea una skill `csv-quality-check` que informe valores ausentes y duplicados sin modificar el CSV.Run the checks before proposing a pull request
In the Andru.ia repository, the skill tells the agent to run npm run chain, npm run validate:references, npm run security:docs and npm run test before proposing the pull request. In another repository, it uses that repository's own current checks instead of inventing commands.
If a check fails, the agent should fix the source and run it again. It should not weaken validators, mark an unrun example as tested, or move toward merging with failures still open. Generated records belong to the canonical sync flow, so they are not edited by hand.
npm run chain
npm run validate:references
npm run security:docs
npm run testSafety notes
The skill runs no scripts and uses no network, according to its description. Review the SKILL.md before installing, as you would any instruction file an agent loads.
The skill itself warns that writing a skill does not prove its examples work. Verify executable examples and any included resources yourself, and ask for expert review where it matters.
- Runs scripts: no
- Uses network: no
- Writing the text does not prove the examples work
What people use it for
- Drafting a new skill for a team's repeatable data-quality check, with defined inputs and outputs
- Improving an existing SKILL.md so its examples and limits are clearer
- Preparing a pull request that adds a skill to the Andru.ia collection
- Writing a checklist-driven SKILL.md for your own project's agent workflow
Frequently asked questions
How do I install Andru.ia Skill-Smith?
Download the ZIP from this page and unzip it. Put the folder in ~/.claude/skills/ for personal use or in a project's .claude/skills/ folder, or upload it in the Claude apps under Settings → Capabilities → Skills.
What does Skill-Smith require?
It requires Git and a local clone of the target repository. For the Andru.ia checks it also needs Node.js and npm, and access to docs/contributors/skill-template.md. It runs no scripts and uses no network.
Is Skill-Smith free?
The skill is published under the MIT licence. This guide does not state any price, so check the directory page for the current terms.
Does Skill-Smith publish or install skills automatically?
No. It only drafts and prepares the work. It does not install or publish anything on its own.
Can I use Skill-Smith outside the Andru.ia repository?
Yes, but its instructions are tailored to Andru.ia and reference its npm scripts and template path. In another repository it is meant to use that repository's own current checks, and you should review the output for mismatched conventions.
What should I do if a check fails?
Fix the cause in the source files and run the check again. Do not weaken the validator, and do not treat an unrun example as tested.
Is Skill-Smith the right tool for using an existing skill?
No. It is for authoring skills, not for tasks such as data checks or document work. Install the skill you actually need instead.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Skill-Smith suits maintainers of the Andru.ia skill collection, developers writing their own SKILL.md skills, and contributors preparing a pull request that adds or changes a skill. If you only want to use an existing skill for a task, or your repository has very different conventions, pick a different tool or adapt the output carefully.
Download the skill