Build, load, and distribute native Lua C/C++/Rust extension modules (shared libraries) for moon_rs: dlopen-based require, resolving the Lua C API from the host executable, Rust cdylib extensions (thrift), C/C++ extensions via the cc crate (crypt, gamecore), the cargo xtask workflow (git + local path mode), and Windows PE import-library generation. Use when creating a new .so/.dylib/.dll extension, debugging a require/symbol-export failure, registering an extension in extensions.toml, or touching xtask / crates/moon-app/build.rs.
Write or modify async Rust↔Lua native modules in moon_rs (lua_<feature>.rs + lualib wrapper): the session-based request/response bridge between a synchronous Lua actor and the Tokio IO runtime, connection/worker pools, graceful shutdown/draining, and the FFI longjmp-safety rules. Use when adding a new native module, adding a Lua-facing C function, wiring DB/network drivers, or touching anything under crates/moon-runtime/src/modules/lua_*.rs.
Review code changes in moon_rs (Rust↔Lua game server framework) against project-specific correctness and safety rules: unsafe Lua FFI patterns, longjmp safety, actor concurrency, session lifecycle invariants, low-copy IO design, skiplist arena soundness, and the accepted patterns that should NOT be flagged. Use when reviewing any PR, commit, or diff in this repo.
Write or modify Lua services (actors) in moon_rs: service skeletons, message dispatch/call/response/send, coroutine flow (moon.async/wait/sleep), the per-service session/PTYPE conventions, service lifecycle and the unique-service shutdown gotcha, and cluster.call/send usage. Use when authoring or editing Lua under lualib/ or assets/ (service scripts, moon.dispatch handlers, cluster scripts), or when a moon_rs process hangs on exit / a coroutine never returns.
Performance profiling and A/B benchmarking for moon_rs Rust↔Lua modules. Covers samply (end-to-end CPU profiling with Lua/FFI flame graphs), pprof (pure-Rust flame graphs via criterion/pprof crate), and A/B comparison scripts (ab_zset.py). Use when diagnosing a performance regression, evaluating an optimization, or profiling any moon_rs native module.