schoolmarm 0.1.1

GBNF grammar-constrained decoding for LLM inference, ported from llama.cpp
Documentation
# Contributing to SchoolMarm


Thank you for your interest in SchoolMarm!

## Open Source, Not Open Contribution


SchoolMarm is **open source** but **not open contribution**.

- The code is freely available under a permissive dual license (MIT OR Apache-2.0)
- You can fork, modify, use, and learn from it without restriction
- **Pull requests are not accepted by default**
- All architectural, roadmap, and merge decisions are made by the project maintainer

This model keeps the project coherent, maintains clear ownership, and ensures consistent quality. It's the same approach used by SQLite and many infrastructure projects.

## How to Contribute


If you believe you can contribute meaningfully to SchoolMarm:

1. **Email the maintainer first**: [michaelallenkuykendall@gmail.com]mailto:michaelallenkuykendall@gmail.com
2. Describe your background and proposed contribution
3. If there is alignment, a scoped collaboration may be discussed privately
4. Only after discussion will PRs be considered

**Unsolicited PRs will be closed without merge.** This isn't personal — it's how this project operates.

## What We Welcome (via email first)


- Bug reports with detailed reproduction steps (Issues are fine)
- Security vulnerability reports (please email directly)
- Documentation improvements (discuss first)
- Compatibility fixes (discuss first)

## What We Handle Internally


- New features and architectural changes
- API design decisions
- Core grammar engine updates
- Performance optimizations

## Bug Reports


Bug reports via GitHub Issues are welcome! Please include:
- Rust version and SchoolMarm version
- OS and version
- Minimal reproduction case
- Expected vs actual behavior
- Grammar file used if relevant