The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill retrieves Markdown content from Jina Reader's official service, which is a well-known tool for converting web pages into a format suitable for AI agents.
[DATA_EXFILTRATION]: The scripts/read_url.py script transmits user-provided target URLs to the Jina Reader API. It also correctly handles the JINA_API_KEY from the environment to provide authenticated access to the service.
[COMMAND_EXECUTION]: The skill executes scripts/read_url.py, which is a local Python script. This script contains functionality to write fetched data to a local file path if the --output flag is provided by the agent or user.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes untrusted content from the public web.
Ingestion points: External content enters the agent context via the output of scripts/read_url.py as described in SKILL.md.
Boundary markers: The SKILL.md instructions guide the agent to inspect the content for completeness and metadata, though the script does not use specific structural delimiters to isolate the untrusted output.
Capability inventory: The agent can perform network requests and write to the local filesystem through the provided Python script.
Sanitization: The script decodes the fetched Markdown content but does not perform additional sanitization or filtering of potential instructions embedded within the source content.

jina-read-url