The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted metadata which could be used as an injection vector.
Ingestion points: The script scripts/build_graph.py reads metadata from user-provided bibliography files and the OpenAlex API.
Boundary markers: No specific delimiters are used in the generated Markdown vault files to distinguish metadata from potential instructions.
Capability inventory: The skill writes various graph data files and Markdown notes to the local system.
Sanitization: The script uses standard formatting but does not sanitize research titles or topics against prompt injection patterns.
[EXTERNAL_DOWNLOADS]: The skill fetches scholarly metadata from a remote source.
Evidence: scripts/build_graph.py queries api.openalex.org for paper details.
Context: OpenAlex is a well-known scholarly metadata repository, and its use is standard for science mapping tasks.

research-knowledge-graph