Guide issue repro, build, and test workflows for TVM FFI changes.
npx skills add https://github.com/guan404ming/gmccc --skill tvm-ffi-dev
Please help me install this Agent Skill. Skill: tvm-ffi-dev Source: https://github.com/guan404ming/gmccc/tree/main/skills/tvm-ffi/tvm-ffi-dev Command: npx skills add https://github.com/guan404ming/gmccc --skill tvm-ffi-dev
Develop and verify TVM FFI changes to streamline debugging, repro and validation workflows.
Run the /tvm-ffi-dev command against an issue URL or description to start the verification loop.
High-intent search queries and answers about installing and using this skill.
To reproduce a TVM FFI issue, you need to create an issue-driven repro script that reliably triggers the bug. This establishes a consistent dev loop to guide your debugging and verify the FFI changes.
The standard workflow for TVM FFI development is a repeatable dev loop. It involves reproducing the issue, building the code, and running validation checks to confirm your changes work reliably.
To validate C++ and Python code style for TVM FFI projects, run standard end-to-end checks during your build process. This ensures consistency and safety across your FFI code changes.
Yes, verifying TVM FFI code requires standard C++ and Python tooling. You need these tools to build the environment, run repro scripts, and validate safety checks effectively.
If your TVM FFI repro script is not working, it may break the consistent dev loop. Ensure your repro script is issue-driven and that your C++ and Python tooling environment is correctly built.
guan404mingExecute CTest and PyTest suites to validate TVM FFI bindings.
guan404mingCompile TVM FFI C++ projects from source using CMake.
guan404mingResolve TODOs and issues in the TVM relax folder with minimal Python repros.
4lermanProfile Node.js, Python, and Go applications for CPU, memory, and I/O bottlenecks.
