The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill makes network requests to https://r.jina.ai/, a well-known service used to read and convert web pages to Markdown. The implementation uses standard Python libraries for network communication.
[COMMAND_EXECUTION]: The skill includes a Python script scripts/read_url.py that handles URL normalization and API interaction. The script allows for writing output to a local file if the --output flag is used, which is consistent with its stated purpose of saving fetched artifacts.
[PROMPT_INJECTION]: The skill operates as an ingestion point for arbitrary public web content, creating an indirect prompt injection surface where malicious instructions on a fetched page could attempt to influence the agent.
Ingestion points: scripts/read_url.py fetches data from any public HTTP/HTTPS URL.
Boundary markers: The fetched content is printed or saved as Markdown without specific delimiters or instructions to the LLM to ignore embedded commands.
Capability inventory: The skill can read network resources and write content to local files.
Sanitization: No sanitization or filtering is applied to the fetched Markdown before it is returned to the agent context.

jina-reader