The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides code snippets for executing external CLI tools using Node.js execFileSync and spawnSync functions, which can run arbitrary system commands if misconfigured.
[PROMPT_INJECTION]: The skill documents an architecture for injecting data into every agent prompt via a context field, creating an indirect prompt injection surface.
Ingestion points: The fileContent parameter in the lsp.check example receives untrusted data from the agent context.
Boundary markers: The provided templates do not include delimiters or instructions for the agent to ignore embedded commands in the injected content.
Capability inventory: The skill body contains examples of execFileSync, spawnSync, fs.writeFileSync, and http.request.
Sanitization: No sanitization or escaping of external content is present in the provided code examples.
[REMOTE_CODE_EXECUTION]: The skill includes a pattern described as "HTTP eval against a running server" using http.request to fetch and process data from local network services.

create-lang-plugin