# Aigent Builder

> >

- **Type:** Skill
- **Install:** `agentstack add skill-wkusnierczyk-aigent-skills-aigent-builder`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [wkusnierczyk](https://agentstack.voostack.com/s/wkusnierczyk)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [wkusnierczyk](https://github.com/wkusnierczyk)
- **Source:** https://github.com/wkusnierczyk/aigent-skills/tree/main/skills/aigent-builder

## Install

```sh
agentstack add skill-wkusnierczyk-aigent-skills-aigent-builder
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# AIgent Builder

Generates complete AI agent skill definitions from natural language purpose descriptions. Produces a valid `SKILL.md` file with YAML frontmatter and structured Markdown body following the [Anthropic agent skill best practices](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/best-practices).

## When to Use

Use this skill when:
- The user asks to create, build, or generate a new agent skill
- The user describes functionality they want to package as a skill
- The user wants to scaffold a new `SKILL.md` file

## Instructions

### Step 1: Detect aigent availability

Run the following to check if the `aigent` CLI is installed:

```bash
command -v aigent
```

### Step 2a: With aigent (primary mode)

If `aigent` is on PATH, use the CLI for authoritative skill generation:

1. **Build the skill:**

   ```bash
   aigent build "" --dir .claude/skills
   ```

   Additional options:
   - `--name ` — override the derived skill name
   - `--license ` — set a license (e.g., `MIT`)
   - `--compatibility ` — set compatibility (e.g., `claude`)
   - `--allowed-tools ` — set allowed tools (e.g., `Bash, Read, Write`)
   - `--no-llm` — force deterministic mode (no API key required)

   If `ANTHROPIC_API_KEY` is set, `aigent build` uses Claude to generate richer names, descriptions, and body content. Without an API key, it falls back to deterministic generation automatically.

2. **Validate the output:**

   ```bash
   aigent validate .claude/skills/
   ```

   Exit code 0 means valid. Any errors are printed to stderr.

3. Report the output directory path and any warnings to the user.

### Step 2b: Without aigent (fallback mode)

If `aigent` is not available, generate the skill manually using the Anthropic spec:

1. **Derive a skill name** from the user's purpose:
   - Convert to kebab-case (lowercase, hyphens between words)
   - Use gerund form for the verb (e.g., "process" becomes "processing")
   - Maximum 64 characters
   - Must not contain reserved words: `anthropic`, `claude`
   - No leading, trailing, or consecutive hyphens

2. **Write the SKILL.md** with this structure:

   ```markdown
   ---
   name: 
   description: . Use when .
   ---
   # 

   

   ## When to Use

   Use this skill when:
   - 
   - 

   ## Instructions

   1. 
   2. 
   3. 
   ```

   Rules for the description:
   - Write in third person ("This skill processes..." not "Process...")
   - Include a "Use when..." clause for auto-invocation triggers
   - Maximum 1024 characters
   - No XML-like tags (``)

3. **Write to** `.claude/skills//SKILL.md` using the Write tool (or a user-specified directory).

4. Inform the user that validation was not run because `aigent` is not installed. Recommend installing it for authoritative validation.

### Output location

The default output directory is `.claude/skills//` (project-local skills). The user may specify an alternative directory. Both CLI mode and fallback mode use the same default.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [wkusnierczyk](https://github.com/wkusnierczyk)
- **Source:** [wkusnierczyk/aigent-skills](https://github.com/wkusnierczyk/aigent-skills)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-wkusnierczyk-aigent-skills-aigent-builder
- Seller: https://agentstack.voostack.com/s/wkusnierczyk
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.