fusion-bench
PassAudited by ClawScan on May 1, 2026.
Overview
This appears to be a benign FusionBench usage guide, with the main caution that it asks users to install and run an external Python package.
This skill looks appropriate if you intend to use FusionBench. Before installing, verify the PyPI package or repository, consider pinning a version, and run the commands in an isolated Python environment because they execute third-party code and may use substantial compute resources.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Installing the package will add and run third-party Python code in the user's environment.
The skill tells users to install an external Python package from PyPI, and the command is not version-pinned. This is purpose-aligned for a FusionBench skill, but it means local execution depends on the external package supply chain.
PyPI: `pip install fusion-bench`
Install in a virtual environment, verify the package source, and pin a known-good version if reproducibility or supply-chain control matters.
Running the examples may execute local Python code, load models, consume compute resources, and create experiment outputs.
The skill provides shell commands to run FusionBench locally. This is central to the skill's stated purpose and is presented as user-directed usage, not hidden automatic execution.
fusion_bench method=task_arithmetic modelpool=clip-vit-base-patch32 taskpool=clip-vit-base-patch32_8tasks
Review commands before running them, use an isolated project environment, and confirm model/config choices before launching large experiments.