llmstxt2skill

Convert llms.txt files to Claude Code Skill format.

Installation

pip install llmstxt2skill

For development:

git clone https://github.com/akuwano/llmstxt2skill.git
cd llmstxt2skill
uv pip install -e ".[dev]"

Usage

# Basic usage
llmstxt2skill https://docs.databricks.com/llms.txt

# Preview without writing (dry-run)
llmstxt2skill https://docs.databricks.com/llms.txt --dry-run

# Custom skill name
llmstxt2skill https://docs.databricks.com/llms.txt --name databricks

# Custom output directory
llmstxt2skill https://docs.databricks.com/llms.txt -o ./skills

# Overwrite existing skill
llmstxt2skill https://docs.databricks.com/llms.txt --force

# Check version
llmstxt2skill --version

Options

LLM Enrichment

Use --enrich to generate high-quality skills with:

• Localized content (Japanese by default)
• Translated and structured documentation links (all links preserved)
• Trigger conditions for when to use the skill
• Capabilities and limitations
• Usage instructions

Note: The --enrich option calls external LLM APIs, which may incur costs depending on your provider and usage.

Supported Providers

Note: The openai-compatible provider is intended for local LLM servers (vLLM, Ollama, llama.cpp) and uses HTTP by default. For production use with remote endpoints, ensure HTTPS is configured.

Examples

# Databricks (default provider)
export DATABRICKS_HOST="https://your-workspace.cloud.databricks.com"
export DATABRICKS_TOKEN="your-token"
llmstxt2skill https://docs.databricks.com/llms.txt --enrich

# OpenAI
export OPENAI_API_KEY="sk-..."
llmstxt2skill https://example.com/llms.txt --enrich --provider openai

# Anthropic
export ANTHROPIC_API_KEY="sk-ant-..."
llmstxt2skill https://example.com/llms.txt --enrich --provider anthropic

# Local LLM (vLLM, Ollama, llama.cpp server)
export OPENAI_BASE_URL="http://localhost:8000"
llmstxt2skill https://example.com/llms.txt --enrich --provider openai-compatible --model llama3

# Specify model explicitly
llmstxt2skill https://example.com/llms.txt --enrich --provider openai --model gpt-4o

# English output
llmstxt2skill https://example.com/llms.txt --enrich --lang en

# Increase max tokens for large llms.txt files
llmstxt2skill https://docs.databricks.com/llms.txt --enrich --max-tokens 20000

Output

By default, skills are written to ~/.claude/skills/{skill-name}/SKILL.md

Use -o or --output to specify a custom output directory:

llmstxt2skill https://example.com/llms.txt -o ./my-skills
# Output: ./my-skills/{skill-name}/SKILL.md

Example

Input (llms.txt):

# Databricks Documentation

> Comprehensive documentation for the Databricks platform.

## Overview
- [Main docs](https://docs.databricks.com/) - How-to guides

Output (SKILL.md):

---
name: databricks-documentation
description: Comprehensive documentation for the Databricks platform.
---

# Databricks Documentation

Comprehensive documentation for the Databricks platform.

## Overview
- [Main docs](https://docs.databricks.com/) - How-to guides

Development

# Install dev dependencies
uv pip install -e ".[dev]"

# Run tests
uv run pytest tests/ -v

# Run linter
uv run ruff check src/ tests/

License

Apache License 2.0 - see LICENSE for details.