The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a bundled Python script (fetch_papers.py) to automate the data collection process. This script is part of the skill package and resides in the local file system.
[EXTERNAL_DOWNLOADS]: The skill performs network requests to fetch academic metadata and RSS feeds from well-known scholarly publishers (e.g., Wiley, Oxford University Press, Elsevier) and the Crossref API.
[DATA_EXFILTRATION]: No evidence of unauthorized data exfiltration was detected. Network operations are strictly limited to retrieving academic content from established domains.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (paper abstracts and titles). While this presents an attack surface for indirect prompt injection, the risk is mitigated by the specific use of reputable academic sources and basic text sanitization (HTML tag removal) performed by the script.
Ingestion points: Data enters the system via the fetch_rss and fetch_crossref functions in fetch_papers.py.
Boundary markers: The skill instructions specify a clear Markdown structure for the final output, which helps the agent distinguish between its own instructions and the processed data.
Capability inventory: The skill utilizes Python execution, network access (GET requests), and local file system write access for maintaining a history log.
Sanitization: The script includes a regex-based cleaner to remove JATS XML and HTML tags from abstracts before they are processed by the agent.

journal-digest