Diagnose Rust crashes, panics, and hangs on host, Android, and iOS targets.
npx skills add https://github.com/po4yka/rust-skills --skill rust-debugging-po4yka
Please help me install this Agent Skill. Skill: rust-debugging Source: https://github.com/po4yka/rust-skills/tree/main/skills/rust-debugging Command: npx skills add https://github.com/po4yka/rust-skills --skill rust-debugging-po4yka
_RNv frames to likely causes and concrete next steps. - Host-first debugging workflow: Drive RUST_BACKTRACE, rust-lldb, rust-gdb, debug-info profiles, and tracing span timing to reproduce crashes locally before touching a device. - Android and iOS device debugging: Filter logcat, read tombstones, symbolicate with ndk-stack, llvm-addr2line, and atos, and attach LLDB from Android Studio or Xcode. - FFI and async diagnosis: Interpret UniFFI panic propagation into Kotlin and Swift, guard extern "C" exports, handle SIGPIPE, and use tokio-console for async stalls. - Use Case: Your Android app crashes with a tombstone showing signal 11 and an address in libmy_ffi.so. Use this Skill to symbolicate the address against the exact unstripped build, identify the dangling pointer, and reproduce it on the host under ASan. ## Quick Start Ask the agent to debug a Rust crash by describing the symptom, such as "my Android app crashes with SIGSEGV in libmy_ffi.so, help me symbolicate the tombstone and find the cause".High-intent search queries and answers about installing and using this skill.
Run the binary or test with RUST_BACKTRACE=1 for a short trace or RUST_BACKTRACE=full for every frame. Set the variable before the process starts, since std caches it at the first capture, and use RUST_LIB_BACKTRACE=0 to disable error-value captures.
Find the first frame in your .so, then run llvm-addr2line -Cfie against the unstripped Cargo output, or pipe the whole tombstone through ndk-stack. The binary must be the exact build that crashed; compare the BuildId with llvm-readelf -n to confirm.
Since Rust 1.81, a panic reaching an extern "C" or extern "system" export aborts the process as defined behavior. Fix it at the export by wrapping the body in catch_unwind or using EnvUnowned::with_env on jni 0.22, not at the caller.
No. The Xcode ASan and TSan switches only instrument code that Xcode compiles, so a prebuilt Rust staticlib is not covered, though allocator interception catches some heap errors. Run sanitizers on a host target for full Rust coverage.
Rust 1.97 made v0 mangling the default, and tools predating v0 support print raw _R names. Upgrade the tool (binutils 2.36+, Linux perf 6.16+) or pipe output through rustfilt; on macOS use c++filt -_.
Use tokio-console with console-subscriber and the tokio_unstable cfg in a local debug build to inspect live task state. A lost-waker warning means poll returned Pending without storing cx.waker().clone(); a never-yielded warning means blocking work needs spawn_blocking.
awfixers-stuffManipulate ELF binaries and static archives with GNU binutils tools.
mazicimertReproduce bugs with failing tests and guide a minimal TDD fix workflow.
po4ykaRun Miri and sanitizers on Rust code to detect undefined behavior and memory errors.
micaelmaltaProfile code and record baseline metrics to identify performance bottlenecks.