The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous curl command templates used to interact with the Hugging Face API. These commands are standard for programmatic data retrieval and management on the platform.
[EXTERNAL_DOWNLOADS]: Data is fetched from huggingface.co and arxiv.org. These are well-known and trusted services for research papers and AI models.
[DATA_EXFILTRATION]: While the skill uses $HF_TOKEN for authenticated requests, it only sends this token to official huggingface.co domains for legitimate API operations such as indexing papers or claiming authorship.
[PROMPT_INJECTION]: As the skill is designed to ingest and analyze external paper content, there is an inherent surface for indirect prompt injection from untrusted paper text.
Ingestion points: Paper content is fetched from huggingface.co/papers/{PAPER_ID}.md.
Boundary markers: None specified in the API interaction instructions.
Capability inventory: The skill utilizes curl for network requests.
Sanitization: No explicit sanitization of the fetched markdown is mentioned, which is common for analysis-focused skills.

hugging-face-papers