The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the drudgereport-pp-cli binary using npx from the @mvanhorn scope or go install from the mvanhorn/printing-press-library GitHub repository. These resources belong to the author of the skill.
[REMOTE_CODE_EXECUTION]: The provided installation methods, specifically npx -y and go install, download and execute code from remote registries during the environment setup process.
[COMMAND_EXECUTION]: The skill drives the drudgereport-pp-cli tool using various subcommands like splash, breaking, and headlines to retrieve data. It also uses which and doctor commands for system verification.
[DATA_EXFILTRATION]: The CLI tool supports an output delivery feature (--deliver webhook:<url>) that can be configured to send news data to an arbitrary external URL. Additionally, a feedback mechanism is described that can POST local logs to a remote endpoint if an environment variable is configured.
[PROMPT_INJECTION]: The skill ingests untrusted news headline content from external sources including drudgereport.com and community RSS feeds, creating a surface for indirect prompt injection.
Ingestion points: Fetches live headlines and RSS items from external domains via the page and feed commands (file: SKILL.md).
Boundary markers: The skill suggests using the --json flag for structured output but lacks explicit instructions to ignore commands embedded within the headlines.
Capability inventory: The agent can execute the CLI tool, write data to local files, and transmit data via webhooks (file: SKILL.md).
Sanitization: There is no description of content sanitization or filtering applied to the fetched news data before it is presented to the agent.

pp-drudgereport