phase.rs
Open-source AI SDLC contribution. Merged contributions to an open-source Magic: The Gathering rules engine, written in Rust, compiled to native and WebAssembly, and held to the game's Comprehensive Rules.
The project
phase.rs is a full Magic: The Gathering game engine. Every card's rules text is parsed into a typed structure and every interaction is checked against the official Comprehensive Rules, so a card either works exactly as printed or it doesn't ship.
The bar is high: idiomatic Rust, reusable building blocks instead of one-off fixes, and a cited rule number for every piece of game logic. Maintainers review every change.
What I contribute
Parser and engine support for cards the engine couldn't play yet: new trigger shapes, conditional riders, and static abilities. Each fix is built for the whole class of cards that share the pattern, not just the one card that exposed it.
Open-source AI SDLC contribution
The project publishes a playbook for contributors who lend it their AI. I run that playbook on a schedule: an agent works overnight on my account, and every change still has to pass the project's own gates and a maintainer's review.
- 01
Pick
Feedback on my open pull requests comes first. Otherwise the agent takes the highest-priority gap from the project's own backlog.
- 02
Plan and review
A plan is written and reviewed by an independent agent before any code exists.
- 03
Build and verify
Implementation, then formatting, lint, tests, and the project's combinator-purity gate against the exact commit.
- 04
Ship or stop
If every gate passes, it opens one pull request. If anything fails, no pull request opens. At most two of mine are open at a time, so maintainers aren't buried.