The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the user to execute npx -y @nomadamas/k-skill@0 instruct delivery-tracking as a required first step to obtain instructions. This pattern downloads and executes code from an external registry under an untrusted NPM scope, allowing for arbitrary code execution. A similar command is provided for updates.
[EXTERNAL_DOWNLOADS]: The skill performs automated network requests via curl to fetch logistics data from external domains including cjlogistics.com and service.epost.go.kr.
[COMMAND_EXECUTION]: The instructions automate complex workflows involving curl for HTTP requests, python3 for data parsing, and shell commands for temporary file management (mktemp) and cookie handling.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted HTML and JSON data from external courier websites.
Ingestion points: Content is fetched via curl from carrier endpoints in instruction.md.
Boundary markers: None present to explicitly warn the agent to ignore embedded instructions in the tracking data.
Capability inventory: The skill uses curl, python3, and local file system access for temporary data storage.
Sanitization: The scripts use basic regex-based HTML tag removal and html.unescape for cleaning, but the ingestion of external data remains a potential attack surface.

delivery-tracking