The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions and metadata to install the retraction-checker-pp-cli and its associated MCP server. This is performed via npx -y @mvanhorn/printing-press-library and go install github.com/mvanhorn/printing-press-library/.... These resources originate from the vendor's organization and are intended for the skill's primary function.
[COMMAND_EXECUTION]: The skill facilitates the execution of a local binary, retraction-checker-pp-cli, using the agent's shell capabilities. It provides various commands (check, scan, watch, etc.) to interact with this binary.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted data through its scan command, which reads .bib files and reading lists provided by users.
Ingestion points: Reading list files and bibliography (.bib) files provided to the scan command.
Boundary markers: The skill encourages the use of the --agent or --json flags to provide structured output to the agent, which acts as a data boundary.
Capability inventory: The skill uses the Read Bash tool and the CLI supports file writes (--deliver file:<path>) and network requests (--deliver webhook:<url>).
Sanitization: The documentation does not specify internal sanitization for the content of the processed bibliography files.
[DATA_EXFILTRATION]: The CLI tool includes a --deliver webhook:<url> parameter that allows the output of the paper check (which may include scholarly data or metadata) to be POSTed to an arbitrary external URL. While designed for integration, this mechanism could be misused to route information to attacker-controlled endpoints.

pp-retraction-checker