The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Zotero libraries, including item metadata, tags, and full-text PDF content (ingestion points identified in read-api.md, tags.md, and full-text.md). There are no explicit boundary markers or instructions provided to the agent to treat these inputs as untrusted data rather than instructions. Successful injection could abuse the skill's capabilities, which include library modifications (write-api.md) and writing files to the host filesystem via the dump method (files-attachments.md).
[METADATA_POISONING]: The skill's documentation in SKILL.md contains a 'Current upstream' date of May 2026, which is in the future relative to the current time. Additionally, the skill-author metadata ('K-Dense Inc.') does not align with the provided author context ('eturkes'), though these may simply be clerical errors rather than intentional deception.

pyzotero