The Agent Skills Directory

[SAFE]: The skill provides standard Python interfaces for scientific research tools focused on protein engineering and analysis.
[EXTERNAL_DOWNLOADS]: The skill documentation describes fetching pre-trained model weights from EvolutionaryScale's official repositories and making legitimate API calls to the Forge platform (forge.evolutionaryscale.ai) for inference services.
[DATA_EXFILTRATION]: No suspicious network activity or data exfiltration patterns were detected. Authentication for the Forge API is handled using standard token placeholders or environment variables as recommended best practices.
[PROMPT_INJECTION]: The instructions are strictly technical and scientific, with no evidence of malicious override patterns, jailbreak attempts, or safety filter bypasses.
[COMMAND_EXECUTION]: Installation instructions utilize standard package managers (uv pip install). There are no instances of arbitrary command execution or suspicious subprocess spawning.
[REMOTE_CODE_EXECUTION]: While the skill loads pre-trained weights (from_pretrained), this is standard behavior for AI model wrappers and is conducted from the vendor's official sources.

esm