The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install tools from the vendor's GitHub repository (github.com/mvanhorn/printing-press-library) using go install and an NPM package (@mvanhorn/printing-press-library) via npx.
[DATA_EXFILTRATION]: The metacritic-pp-cli tool supports a --deliver webhook:<url> flag, which enables the transmission of command output to external URLs via HTTP POST. While intended for automation, this feature could be leveraged to exfiltrate data from the environment to an unverified third-party endpoint.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from an external API, creating a vulnerability surface.
Ingestion points: Data enters the agent context through the finder and composer commands which query the Metacritic JSON API (SKILL.md).
Boundary markers: The instructions do not specify the use of delimiters or specific warnings to ignore instructions embedded within the API responses.
Capability inventory: The skill is allowed the Read Bash tool and the underlying CLI binary has capabilities for network transmission (webhooks) and local file writing (--deliver file:<path>).
Sanitization: There is no evidence of sanitization or filtering of the API data before it is presented to the agent.

pp-metacritic