The Agent Skills Directory

[DATA_EXFILTRATION]: The worten-pp-cli tool supports a --deliver webhook:<url> flag, allowing the agent to POST command results to any external URL. This is a direct capability for data exfiltration that can be abused to transmit sensitive information or project context to attacker-controlled servers.
[DATA_EXFILTRATION]: The skill describes a feedback mechanism (worten-pp-cli feedback) that can read data from standard input and potentially transmit it to a remote endpoint if the WORTEN_FEEDBACK_ENDPOINT environment variable is configured. This provides an additional path for sending local data to external locations.
[COMMAND_EXECUTION]: The skill is designed to drive the worten-pp-cli binary by executing shell commands. The instructions explicitly describe passing $ARGUMENTS into shell commands, which presents a risk of command injection if the arguments are not properly sanitized by the binary.
[EXTERNAL_DOWNLOADS]: The skill instructions require the installation of a CLI tool via npx -y @mvanhorn/printing-press-library install worten --cli-only. This involves downloading and executing remote code from the NPM registry at runtime. This resource belongs to the skill's author context but remains a remote dependency.
[PROMPT_INJECTION]: The skill fetches content from the Worten API, including product descriptions, technical specifications, and search suggestions. This data is ingested into the agent's context without sanitization, creating an attack surface for indirect prompt injection where malicious instructions hidden in product data could influence the agent's behavior.
Ingestion points: get-product-details, get-search-suggestions, get-technical-specifications, and search-products commands (SKILL.md).
Boundary markers: None present; API data is interpolated directly into the context.
Capability inventory: The agent can execute shell commands and perform network POST requests via the webhook sink.
Sanitization: No filtering or validation of the external API content is documented.

pp-worten