The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches the arxiv-pp-cli tool using npx or go install from the vendor's infrastructure at github.com/mvanhorn/printing-press-library and the @mvanhorn NPM registry.
[COMMAND_EXECUTION]: The skill operates by executing the arxiv-pp-cli binary with various flags, including a natural-language mapping tool (which) and a diagnostic tool (doctor).
[INDIRECT_PROMPT_INJECTION]: The skill processes research paper metadata from the public arXiv Atom API, which could contain untrusted data.
Ingestion points: The query command fetches external data (titles, abstracts, metadata) from the arXiv API into the agent context via SKILL.md.
Boundary markers: Output is delimited within a structured JSON envelope using a results field.
Capability inventory: The tool includes shell command execution, atomic file writes via file:<path>, and network POST requests via webhook:<url> delivery sinks.
Sanitization: No explicit sanitization or filtering of the API-returned content is documented.

pp-arxiv