Sheave 🧭

Presets for guiding agentic AI workflows.
Because tasks shouldn’t get lost in translation.

Sheave provides a curated collection of AI guidance presets that you can selectively enable for your IDE integrations. Similar to how ruff lets you choose which linting rules to enable, Sheave lets you pick and choose which rules, workflows, and commands to activate.

Quick Start

Install Sheave:

# Using poetry
poetry add sheave

# Using pip
pip install sheave

Enable presets in your project:

sheave enable --rules code-quality --workflows testing
sheave sync

Key Features

• Selective presets — Choose only the rules, workflows, and commands you need
• IDE integration — Works with Cursor, Claude Desktop, and similar tools
• Ruff-like interface — Familiar select and ignore configuration model
• Zero dependencies — Lightweight and focused
• Modular — Enable or disable presets independently
• Configurable — Customize presets to match your project’s needs

What are Presets?

Sheave offers three types of presets:

• Preset Rules: Pre-configured prompt rules that get added to each AI interaction
  • Code quality standards
  • Testing best practices
  • Documentation guidelines
  • Security considerations
• Preset Workflows: Common workflows you can point an AI assistant to
  • Setting up new features
  • Refactoring patterns
  • Debugging strategies
  • Code review checklists
• Preset Commands: Ready-to-use commands for common development tasks
  • Generate test files
  • Create documentation
  • Run code quality checks
  • Format and lint code

Documentation

• Getting Started — Installation and first steps
• Configuration — How to enable and configure presets
• CLI Reference — Command-line options and usage
• API Documentation — Programmatic API for integrations
• Examples — Real-world usage examples

License

MIT-aNOAI License

You’re free to use, copy, and modify the library under the standard MIT terms.
The additional rider simply requests that this project not be used to train or fine-tune AI/ML systems until the author deems fair compensation frameworks exist.
Normal use, packaging, and redistribution for human developers are unaffected.

_{😐 Apathetic Tools © MIT-aNOAI}