Validates code changes against zo's Three Prime Directives and architectural mandates. Use when reviewing PRs, planning new features, or when user says "check architecture", "validate design", "does this follow the rules", "architecture review", or "guard check". Do NOT use for simple formatting, typo fixes, or non-compiler code changes.
Creates or updates a zo changelog file. Use when user says "changelog", "changelogs", "log changes", "update changelog", or after a commit that introduces notable changes. Do NOT use for trivial changes like typo fixes or formatting.
Runs the zo build pipeline (cargo build, clippy, tests) and analyzes failures with targeted fix suggestions. Use when user says "build", "check", "does it compile", "run clippy", "run tests", "ci check", or after making code changes. Do NOT use for non-Rust files or documentation-only changes.
Runs zo's test suite and intelligently analyzes failures. Reads failing test sources, traces the failure to compiler code, and suggests fixes. Use when user says "run tests", "what's failing", "test analysis", "fix failing tests", or "why is this test broken". Do NOT use for benchmarks or performance testing (use zo-perf-bench instead).
Reviews SIR-to-machine-code generation output for correctness and efficiency. Use when user says "review codegen", "check generated code", "codegen quality", "is the output correct", "review emitter output", or "arm output review". Do NOT use for source-level code review (use /blame instead) or performance benchmarking (use zo-perf-bench).
Benchmarks zo compiler phases against performance targets (10M LoC/s parse, 5M LoC/s analyze, 5M LoC/s codegen). Use when user says "benchmark", "perf", "how fast", "performance test", "speed check", "are we hitting targets", or "LoC/s". Do NOT use for general profiling of non-compiler code.