segment-anything-model
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads source code and model weights from official repositories and domains belonging to Facebook Research (Meta), which is a well-known and trusted organization. Examples include
pip install git+https://github.com/facebookresearch/segment-anything.gitandwgetcommands targetingdl.fbaipublicfiles.com. These are standard practices for utilizing large machine learning models. - [INDIRECT_PROMPT_INJECTION]: As a multimodal computer vision skill, it ingests external image data for processing. While this presents an attack surface for adversarial inputs, the skill's capabilities are limited to image manipulation and segmentation within a local environment or isolated service example. There are no patterns suggesting the tool's output is used to drive sensitive shell commands or perform data exfiltration.
- [NO_CODE]: The skill primarily consists of documentation and Python code snippets designed to be run by the user or an agent. It does not include pre-compiled binaries or hidden scripts.
Audit Metadata