The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines patterns for processing external inputs like URLs through the routeToPage(url) function, which creates a surface where an agent could be exposed to instructions embedded in the data.
Ingestion points: The routing logic described in references/examples.md and references/getting-started.md involves processing a url parameter.
Boundary markers: The provided code snippets do not include delimiters or specific instructions to disregard embedded commands in the processed data.
Capability inventory: The skill is restricted to documentation and code examples; it does not include any tools or scripts that perform sensitive actions such as file system writing, network requests, or dynamic code execution.
Sanitization: No sanitization or input validation logic is present in the reference examples.
[EXTERNAL_DOWNLOADS]: The documentation mentions project initialization using pnpm create arrow-js@latest, which triggers a standard framework scaffolding process through the package manager.

arrow-js