Skip to main content

​
Overview

The fastskill init command creates a skill-project.toml file for skill authors. This command extracts metadata from an existing SKILL.md file (if present) and interactively prompts for any missing information to create a properly configured skill project.
This command is for skill authors who want to create a new skill project. It generates the skill-project.toml manifest file that defines your skill’s metadata, dependencies, and configuration. For general FastSkill setup, see the Installation Guide.
Context Detection: fastskill init automatically detects whether you’re working in a skill directory or project root:
  • Skill-level: If run in a directory containing SKILL.md, creates skill-project.toml with [metadata] section (for skill authors)
  • Project-level: If run at project root, can create skill-project.toml with [dependencies] section (for skill consumers)
See Skill Commands for details on skill-level initialization.

​
What It Does

When you run fastskill init, it performs the following steps:
  1. Checks for Existing Files: Validates if skill-project.toml already exists
  2. Reads SKILL.md: Extracts metadata from frontmatter (name, description, version, etc.)
  3. Interactive Prompts: Asks for missing information or uses provided CLI arguments
  4. Creates skill-project.toml: Generates the manifest file with proper structure
  5. Validates Configuration: Ensures all required fields are present and valid

​
Usage

fastskill init
This will prompt you for each configuration option: 
🚀 FastSkill Project Initialization

🔑 OpenAI API Key
Enter your OpenAI API key (or press Enter to set via environment variable): sk-...

✅ API key format validated

🤖 Embedding Model
Choose an embedding model:
  1. text-embedding-3-small (recommended, cost-effective)
  2. text-embedding-3-large (higher accuracy, more expensive)
  3. Custom model
Enter choice (1-3) [1]: 1

📁 Skills Directory
Enter skills directory path [.skills]: .skills

✅ FastSkill project initialized successfully!

💡 Next steps:
  • Add skills to the '.skills' directory
  • Run: fastskill reindex
  • Then: fastskill search "your query"

​
Non-Interactive Setup

# Use defaults for everything (requires OPENAI_API_KEY env var)
fastskill init --yes

# Specify custom options
fastskill init --yes --skills-dir my-skills --embedding-model text-embedding-3-large

# Skip creating example skill
fastskill init --yes --no-example

​
Command Options

OptionDescriptionDefault
--yesSkip interactive prompts, use defaultsfalse
--forceReinitialize even if files existfalse
--skills-dir <DIR>Skills directory path.skills
--embedding-model <MODEL>OpenAI embedding modeltext-embedding-3-small
--no-exampleSkip creating example skillfalse

​
Files Created

​
.fastskill/config.yaml

Service-level configuration file with your embedding settings: 
# FastSkill Configuration
# This file configures embedding search settings

embedding:
  openai_base_url: "https://api.openai.com/v1"
  embedding_model: "text-embedding-3-small"
Note: Service-level configuration (embedding settings) is stored in .fastskill/config.yaml. Project-level configuration (skill dependencies and repositories) is stored in skill-project.toml at your project root.

​
.skills/ Directory Structure

.skills/
├── .fastskill/          # Internal FastSkill data (index, cache)
├── examples/            # Example skills for learning
│   └── hello-world/
│       └── SKILL.md    # Sample skill definition
└── README.md           # Directory documentation

​
Example Skill

Creates a basic skill at .skills/examples/hello-world/SKILL.md: 
---
name: Hello World
description: A simple example skill that demonstrates the FastSkill format
version: 1.0.0
author: FastSkill
tags: [example, demo, hello-world]
capabilities: [text_processing, greeting]
---

# Hello World Skill

This is an example skill to demonstrate the FastSkill format and structure.

## Purpose

This skill demonstrates how to create a basic FastSkill skill definition.

​
Configuration Options

​
Embedding Models

ModelDescriptionCostUse Case
text-embedding-3-smallCost-effective, good performanceLowMost projects
text-embedding-3-largeHigher accuracy, more expensiveHighEnterprise, high-precision needs
CustomAny OpenAI-compatible modelVariesAdvanced users

​
Skills Directory

  • Default: .skills (recommended)
  • Custom: Any directory path you specify
  • Existing: Will be created if it doesn’t exist

​
Environment Variables

​
OPENAI_API_KEY

The init command will use this environment variable if set: 
export OPENAI_API_KEY="your-api-key-here"
fastskill init --yes  # Will use the env var automatically
If not set, you’ll be prompted to enter it interactively.

​
Troubleshooting

​
”API key should start with ‘sk-’”

Cause: The provided API key doesn’t match OpenAI’s format Solution: Check that you’re using a valid OpenAI API key

​
”API key appears to be too short”

Cause: The API key is incomplete Solution: Ensure you’re copying the complete API key from OpenAI

​
”FastSkill is already initialized”

Cause: Configuration files already exist Solution: Use --force to reinitialize, or remove existing files first

​
Permission Denied

Cause: Cannot write to the current directory Solution: Check file permissions or run in a directory where you have write access

​
Integration with CI/CD

For automated setups in CI/CD pipelines: 
# .github/workflows/setup-fastskill.yml
name: Setup FastSkill
on: [push]

jobs:
  setup:
    runs-on: ubuntu-latest
    steps:
      - name: Install FastSkill
        run: cargo install fastskill

      - name: Initialize FastSkill
        env:
          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
        run: fastskill init --yes --no-example

      - name: Add skills
        run: cp -r my-skills/* .skills/

      - name: Index skills
        run: fastskill reindex

​
Best Practices

​
Project Structure

Keep FastSkill configuration at the project root: 
my-project/
├── .fastskill/
│   └── config.yaml  # Service-level configuration (embedding settings)
├── skill-project.toml  # Project-level configuration (dependencies, repositories)
├── skills.lock       # Lock file (auto-generated)
├── .claude/
│   └── skills/      # Skills directory (default location)
│       ├── my-skill/
│       └── another-skill/
└── src/              # Your application code

​
API Key Management

  • Development: Use environment variables
  • Production: Use secure secret management
  • CI/CD: Store in encrypted secrets

​
Version Control

Consider what to commit: 
# Commit these files
git add .fastskill/config.yaml
git add skill-project.toml
git add skills.lock
git add .claude/skills/README.md
git add .claude/skills/examples/

# Don't commit these (generated)
echo ".skills/.fastskill/" >> .gitignore

​
Next Steps After Init

  1. Add Your Skills: Create skill directories under .skills/
  2. Index Skills: Run fastskill reindex to build the search index
  3. Test Search: Try fastskill search "your query"
  4. Integration: Connect to your AI agent or application
The init command gets you started quickly with a properly configured FastSkill environment. All the complexity of configuration files and directory structure is handled automatically.