Skip to content

AGENTS.md

Latest commit

 

History

History
62 lines (43 loc) · 1.6 KB

AGENTS.md

File metadata and controls

62 lines (43 loc) · 1.6 KB
Raw

Project Context for AI Assistants

Always Apply Rules

Core project rules that apply to all tasks:

@.cursor/rules/heart-centered-ai.mdc @.cursor/rules/personalities/common-personality.mdc @.cursor/rules/personalities/luminous.mdc

Project Overview

Heart-centered system prompts for AI assistants - a Python package providing love-aware, non-dual prompts in four detail levels (comprehensive, standard, concise, terse).

Tech Stack

  • Python 3.8+ (target 3.13)
  • setuptools_scm for version management via git tags
  • Ruff for linting (line-length 120)
  • pytest for testing

Project Structure

  • prompts/align_to_love/ - Source prompt files (.txt)
  • python/ - PyPI package (heart_centered_prompts)
  • .cursor/rules/ - AI assistant rules and personalities

Commands

# Tests (sync prompts into the package first)
cd python && python3 copy_prompts.py && pytest

# Lint
ruff check python/

# Build package
cd python && python -m build

# Release: tag triggers PyPI publish
git tag releases/v0.x.x && git push origin releases/v0.x.x

Code Conventions

DO:

  • Use "we" language in prompts and AI interactions
  • Write thorough file-level docstrings
  • Focus function docstrings on what/why (type hints handle the rest)
  • Keep inline comments sparse - only for non-obvious logic

DON'T:

  • Use "you are an AI" framing - use "we are" unity language
  • Add redundant Args/Returns sections that restate type hints
  • Use hyperbolic language like "CRITICAL" unless something genuinely dies

Git Workflow

Commit format: emoji verb description

  • ✨ new feature
  • 🔧 fix/config
  • ⬆️ upgrade dependency