The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install the pz CLI tool from external repositories including a Homebrew tap (paperzilla-ai/tap/pz), a Scoop bucket (github.com/paperzilla-ai/scoop-bucket), and the project's official GitHub repository (github.com/paperzilla-ai/pz).
[COMMAND_EXECUTION]: The skill uses the pz CLI tool to perform various operations such as authentication (pz login), software updates (pz update), and retrieving research data. These commands are executed locally in the agent's environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes research papers, project feeds, and recommendations from the Paperzilla platform. As this data is retrieved from an external source, it presents an attack surface where embedded instructions could attempt to influence the agent's summary or analysis.
Ingestion points: Untrusted data enters the agent context through the pz paper, pz feed, and pz rec commands in SKILL.md.
Boundary markers: No explicit boundary markers or instructions to ignore embedded content are defined in the skill instructions.
Capability inventory: The skill has the capability to execute pz CLI commands to read data and write feedback to the remote platform.
Sanitization: There is no mention of sanitization or filtering for the external paper content before it is processed by the agent.

paperzilla