The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install external executable binaries using go install from a GitHub repository (github.com/mvanhorn/printing-press-library) and via npx from the NPM registry (@mvanhorn/printing-press). These resources originate from the skill's author/vendor.
[COMMAND_EXECUTION]: The skill is designed to drive a local binary (trigger-dev-pp-cli) via shell commands to perform various management tasks. It requests the Read Bash tool to execute these operations, including real-time monitoring and database syncing.
[DATA_EXFILTRATION]: The CLI tool provides a built-in --deliver webhook:<url> flag that allows the agent to POST command output (which may contain sensitive information like environment variable diffs or run data) to an arbitrary external URL. This represents a potential data exfiltration vector if not carefully managed.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by interpolating user-provided strings into shell commands for searching runs, querying capabilities, and recording feedback.
Ingestion points: User queries for the runs find command, the which discovery command, and text provided to the feedback command.
Boundary markers: None identified; user input is passed directly as arguments to the CLI tool.
Capability inventory: The skill uses Read Bash to execute shell commands with the interpolated input.
Sanitization: No explicit sanitization or escaping of the user-provided arguments is described in the instructions.

pp-trigger-dev