The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes the sentence-transformers library to download the all-MiniLM-L6-v2 embedding model from Hugging Face. This is a standard practice for text-based AI skills and relies on a well-known repository for machine learning models.
[COMMAND_EXECUTION]: The provided Python script resources/cluster.py performs clustering using established algorithms (UMAP and HDBSCAN). It operates on local data provided via command-line arguments and does not execute unauthorized shell commands or access sensitive system information.
[PROMPT_INJECTION]: The skill handles event abstracts which are external data. However, it incorporates a safety-by-design approach where the LLM is only used to generate labels for clusters based on a few representative examples, rather than processing every item, which limits the potential for indirect prompt injection to influence the agent's behavior.
[DATA_EXFILTRATION]: No evidence of unauthorized data transmission was found. The skill reads from and writes to the local file system as part of its primary data processing function.

conf-theme-clustering