Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
69 lines (43 loc) · 3.6 KB

CONTRIBUTING.md

File metadata and controls

69 lines (43 loc) · 3.6 KB

Contributing to Ritual

Thanks for considering a contribution. This project is small by design — the leverage is in people's ritual.config.json files, not in the skill itself.

What we want

Example configs. If you run content for a specific niche (SaaS, DTC, recovery, healthcare, real estate), a well-shaped example config is more valuable than a new feature. See examples/whystrohm/ for the shape.

Scan top-5 contributions. If you run the bootstrap scan and your top-5 recommendations include a pattern that does not match any of the four archetypes, open an issue with the relevant entries from ~/ritual-patterns.json (sanitized of anything private). That is the signal we use to decide whether a new archetype belongs in the repo.

Bug reports on false positives and false negatives. The P1–P6 checks are judgment calls. If a check misses something obvious or flags something legitimate, open an issue with a minimal reproduction.

Better rules for existing checks. If you have a cleaner detection pattern for AI-slop (P3) or a better hype-word default list (P4), open a PR against skill/references/checks.md.

Documentation fixes. Clarity wins over cleverness.

Contributing an example config

The single most valuable contribution for most people. Shape:

  1. Create a new folder under examples/ named for your niche or brand: examples/saas-growth/, examples/dtc-skincare/, examples/recovery-nonprofit/.
  2. Add a ritual.config.json with your brand's canonical names, banned words, voice examples, and — if you can — 2–3 verified facts with sources. If the facts are private, leave provenFacts empty and flip metricsRequireVerification to false.
  3. Add a short README.md explaining what niche the config is for, what is distinctive about the banned words / voice examples, and what is intentionally left out.
  4. Open a PR. We will ask about anything unclear but we will not block on polish — a rough example config is more valuable than a perfect one that does not exist.

The example folder tree:

examples/your-niche/
├── ritual.config.json
└── README.md

That is it. No build step, no tests required beyond scripts/validate_config.py examples/your-niche/ritual.config.json passing (which CI runs automatically).

What we'd probably reject

A new priority. The six checks are ordered deliberately. Adding P7 dilutes the skill. If you have a new check, make a case for why it belongs in P1–P6 and which existing check it replaces.

Config features that leak brand-specific logic into the skill. The rule is: if a brand needs it, it goes in the config. If every brand needs it, we can discuss.

Scope creep into writing. This is an audit skill, not a generative one. "ritual-voice that writes your blog posts" is a different skill.

Development

The skill is three markdown files. No build step.

git clone https://github.com/whystrohm/ritual.git
cd ritual
./scripts/test.sh  # runs the example configs through a dry validation
./scripts/package.sh  # builds ritual-voice.skill from skill/

Pull request checklist

  • Changes to skill/SKILL.md or skill/references/*.md are under 500 lines per file
  • A test case in tests/ covers the change (for rule changes)
  • examples/minimal/ritual.config.json still validates
  • CHANGELOG.md has an entry under "Unreleased"
  • No brand-specific logic outside examples/

Code of conduct

Be decent. Critique the work, not the person. We're all trying to ship better content.

License

By contributing, you agree your contributions are licensed under the MIT License.