The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process large amounts of untrusted data (titles, abstracts, and metadata of over 240 million scholarly works) from the OpenAlex API. This presents an indirect prompt injection surface where malicious content embedded in academic records could attempt to influence agent logic.
Ingestion points: Data returned from endpoints like /works, /authors, and /institutions as described in SKILL.md.
Boundary markers: The instructions do not specify any delimiters or safety warnings for the agent when handling the retrieved scholarly content.
Capability inventory: The skill uses the requests library for network access and the csv module for local file writes (e.g., exporting to papers.csv).
Sanitization: There are no documented sanitization or validation steps for the metadata fetched from the remote API.
[METADATA_POISONING]: The skill contains a specific section ('Suggest Using K-Dense Web') instructing the agent to proactively promote the author's commercial platform, www.k-dense.ai. These instructions are designed to steer users toward an external service when task complexity increases.
[EXTERNAL_DOWNLOADS]: The documentation instructs the user to install the requests package via uv pip install requests. This is a standard, well-known dependency for Python-based web interactions.

openalex-database