The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses curl to interact with official Italian government APIs and python3 to parse and filter JSON data. These commands are used for legitimate data processing aligned with the skill's primary purpose.
[EXTERNAL_DOWNLOADS]: The skill fetches data from www.indicepa.gov.it and indicepa.gov.it. These are official domains of the Italian government (AgID) and are recognized as trusted sources for this specific use case.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external API responses. While this represents a theoretical attack surface, the skill only performs structured JSON parsing and attribute extraction, which significantly limits the risk of the agent interpreting data as instructions.
[CREDENTIALS_UNSAFE]: The skill utilizes an environment variable IPA_auth_id for authentication. It includes critical instructions for the agent to verify the variable's presence without ever revealing its value in output, logs, or command execution strings.

ipa