The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches bibliometric metadata and paper details from well-known academic services including OpenAlex, Scopus, Web of Science, and Crossref.- [EXTERNAL_DOWNLOADS]: Downloads preprint LaTeX source archives from arXiv, which is a trusted academic repository.- [COMMAND_EXECUTION]: Executes system commands using curl, wget, tar, and uv to facilitate data retrieval from APIs, extraction of source archives, and execution of local Python packages.- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface due to the ingestion and processing of untrusted academic data.
Ingestion points: Research paper abstracts, bibliometric metadata, and LaTeX source files retrieved from external web sources (SKILL.md, references/agent-templates.md, references/cli-council-search.md).
Boundary markers: Absent. There are no explicit delimiters or 'ignore' instructions wrapping the external content when it is passed to the agent context.
Capability inventory: The agent can execute shell commands via Bash (curl, wget, tar, uv), perform network operations, and write files (SKILL.md, references/cli-council-search.md).
Sanitization: Not identified. The skill relies on the LLM to interpret retrieved academic content without explicit filtering or validation against prompt injection patterns.

literature