pwasm 0.2a0 has been released. The project is an experimental, pure-Python WebAssembly engine the author started in January as a personal "folly project." After not touching it for months, they turned Claude Opus 5.5 loose on the code to evaluate its current state and explore what would be needed to run the MicroPython and micro JavaScript experiments from the research repository under pwasm, as well as to identify opportunities to speed it up.
Goals and process
The request to the model covered three main points: assess pwasm's existing functionality, determine the work required to support MicroPython and the micro JavaScript experiments, and suggest ways to improve performance. Following that intervention and with minimal follow-up prompting, the repository received 42 commits.
Results
- After the updates, pwasm now implements almost the entire WASM specification.
- The PyPI wheel produced by the project bundles working WASM builds of MicroPython, QuickJS and Micro QuickJS.
Caveats
The release is labeled alpha and the author explicitly notes they would not trust it for production or critical use. The development serves as a demonstration of how current large language models can improve on work done by models from about ten months earlier, but it remains an experimental codebase.
Tags
python, webassembly, vibe-coding
Summary
pwasm 0.2a0 documents a significant iterative improvement driven by Claude Opus 5.5: 42 commits yielded near-complete WASM support and packaged builds for several embedded runtimes, while the project still carries an alpha warning and should be treated as experimental.
