Background Removal
PassAudited by ClawScan on May 1, 2026.
Overview
This instruction-only skill is coherent for background removal, but users should notice that it depends on a third-party CLI, account login, and external image-processing calls.
Before installing, verify the inference.sh CLI installer or use the manual checksum flow, log in only with the account you intend to use, and avoid sending private images or URLs unless you are comfortable with third-party processing.
Findings (4)
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.
Running the setup command would execute code downloaded from inference.sh before using the skill.
The skill documents installing the required CLI by piping a remote script into a shell. It is user-directed and central to the stated purpose, but the runnable installer is not included in the artifact set.
curl -fsSL https://cli.inference.sh | sh && infsh login
Review the installer or use the linked manual install/checksum process before running it, especially on sensitive machines.
The agent may run inference.sh commands under the account that is logged in on the machine.
The workflow expects the user to authenticate to inference.sh, although the registry metadata lists no primary credential. This is expected for a cloud CLI but still means actions may use the user's account.
infsh login
Use the intended inference.sh account and understand any billing, quota, or workspace implications before invoking the skill.
The agent has latitude to invoke different inference.sh CLI actions, not only the single BiRefNet background-removal example.
The skill allows Bash use for any command beginning with infsh, while the examples include multiple apps such as infsh/birefnet, falai/reve, and falai/flux-dev-lora. This is aligned with the inference.sh workflow but broader than one background-removal command.
allowed-tools: Bash(infsh *)
Ask the agent to confirm the exact infsh app and input image before running commands if cost, privacy, or model choice matters.
Image URLs and related prompts may be processed by third-party inference services.
The documented workflow sends an image URL to an external inference service. This is inherent to the skill's purpose, but it affects privacy for images or URLs that are not meant to be shared.
infsh app run infsh/birefnet --input '{"image_url": "https://your-photo.jpg"}'Use only images you are comfortable sending to the selected provider, and avoid private or restricted URLs unless you understand the provider's data handling.