The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the user or agent to run npx -y @nomadamas/k-skill@0 instruct naver-news-search as a required first step to fetch primary instructions. This involves downloading and executing code from the NPM registry at runtime, making the skill's logic dependent on a remote, mutable package.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the Naver News Search API via an external proxy.
Ingestion points: External data enters the context through items[].title and items[].description fields retrieved from the proxy response as described in instruction.md.
Boundary markers: The instructions guide the agent to quote the API description directly and attribute the information specifically to the article summary ("기사 요약에 따르면"), which acts as a basic boundary to prevent the AI from adopting instructions within the news text.
Capability inventory: The skill has the capability to execute shell commands (specifically curl) to interact with the search proxy.
Sanitization: The proxy server is documented to remove HTML tags like <b> and standard HTML entities before passing data to the agent, reducing the risk of markup-based injection.
[COMMAND_EXECUTION]: The skill utilizes shell commands (curl) to fetch data from the search proxy and suggests using npx for setup and documentation retrieval.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to the vendor-controlled proxy at https://k-skill-proxy.nomadamas.org/v1/naver-news/search to retrieve search results.

naver-news-search