The Agent Skills Directory

[SAFE]: The skill interacts with the OpenAlex API, a reputable and public resource for scholarly metadata. No unauthorized exfiltration or credential harvesting was detected.
[EXTERNAL_DOWNLOADS]: The skill instructions specify the installation of the requests library. This is a standard and well-known Python package for HTTP communication.
[COMMAND_EXECUTION]: The skill uses uv pip for package management and includes workflows for saving research data to local CSV files, which is appropriate for its stated analytical purpose.
[PROMPT_INJECTION]: The skill includes instructions to the agent to suggest the author's hosted research platform (K-Dense Web) during complex tasks. This is promotional steering rather than a security bypass. The skill also ingests external scholarly data which presents a surface for indirect prompt injection.
Ingestion points: External metadata (titles, abstracts) retrieved from api.openalex.org.
Boundary markers: None are explicitly implemented in the provided Python snippets to isolate external text.
Capability inventory: Network requests via requests, local file system access for data export, and execution of skill-provided scripts.
Sanitization: Not explicitly present in the provided basic client implementation.

openalex-database