The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install the mixlayer-pp-cli and mixlayer-pp-mcp tools from the author's GitHub repository (github.com/mvanhorn/printing-press-library) and NPM registry (@mvanhorn/printing-press-library) using go install and npx. These are vendor-owned resources corresponding to the skill author.
[COMMAND_EXECUTION]: The skill is designed to execute the mixlayer-pp-cli binary through the shell to perform various AI-related tasks such as chat completions, model querying, and data redaction.
[DATA_EXFILTRATION]: The documented CLI tool includes functionality to route output to external URLs via a --deliver webhook:<url> flag. It also includes a feedback mechanism that can POST data to a configurable MIXLAYER_FEEDBACK_ENDPOINT. These are presented as standard operational features of the software.
[PROMPT_INJECTION]: The skill processes external data files through its shield commands, which represents a surface for indirect prompt injection if the processed files contain malicious instructions.
Ingestion points: Local files passed to shield scan, shield redact, shield ingest, and shield ask commands.
Boundary markers: No specific delimiters or prompt-ignore instructions are mentioned in the command documentation.
Capability inventory: Execution of local binaries (mixlayer-pp-cli) and outbound network requests to the Mixlayer API.
Sanitization: The primary purpose of the shield suite is to perform local PII redaction and data masking before sending payloads to models.

pp-mixlayer