The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of a specialized CLI tool from external sources not listed in the trusted vendors repository. It specifically instructs the agent to run npx -y @mvanhorn/printing-press-library to install the ars-sicilia tool. It also provides a fallback installation method using go install targeting the github.com/mvanhorn/printing-press-library repository. While these resources are consistent with the identified vendor infrastructure ('mvanhorn'), they represent the execution of remote code from non-whitelisted sources.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process unstructured data from external public web portals (dati.ars.sicilia.it and www.ars.sicilia.it). This creates a surface for indirect prompt injection if those portals contain malicious content meant to influence the agent's behavior.
Ingestion points: Data is fetched from the ARS Sicilia portal, including bill titles, parliamentary minutes (resoconti), and commission summaries (SKILL.md).
Boundary markers: The instructions do not define clear delimiters or "ignore" directives to prevent the agent from following instructions potentially embedded in the fetched documents.
Capability inventory: The skill requires Read Bash tool access and executes shell commands via the ars-sicilia-pp-cli binary (SKILL.md).
Sanitization: There is no mention of sanitization, filtering, or validation of the remote data before it is presented to the agent's context.
[COMMAND_EXECUTION]: The skill heavily relies on the execution of the ars-sicilia-pp-cli binary to perform its core functions. It also includes a --deliver webhook:<url> feature that allows the tool to POST its output to a remote URL, which could be leveraged for data exfiltration if the agent is directed to an attacker-controlled endpoint.

pp-ars-sicilia