The Agent Skills Directory

[DATA_EXFILTRATION]: The CLI tool accesses the user's ~/.env file to read the OPENALEX_API_KEY. This key is then transmitted in the query string to api.openalex.org and content.openalex.org for authentication. While this is standard practice for API clients, it involves reading sensitive local configuration and transmitting it to a remote service.
[EXTERNAL_DOWNLOADS]: The download-pdf command retrieves files from external URLs provided by the OpenAlex API. The script includes a safety check to verify the %PDF- header before writing to disk, but it still involves downloading content from arbitrary scholarly hosting domains.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill ingests untrusted scholarly metadata (such as titles and abstracts) from the OpenAlex API via scripts/openalex_cli.py.
Boundary markers: There are no explicit boundary markers or instructions to the agent to ignore embedded instructions in the retrieved data, although the skill body suggests using jq to reduce context size.
Capability inventory: The skill possesses the capability to write files to the local filesystem (via the download-pdf command) and performs network operations.
Sanitization: The script parses JSON data but does not sanitize text content (like abstracts) for malicious prompt instructions before presenting it to the agent.

literature-search-openalex