The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation provides instructions for connecting to a remote MCP server and installing the skill from https://unbrowse.ai. These resources are hosted by the skill's vendor and are used to provide the primary functionality of searching and fetching content from radiofrance.fr.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of data from an external source (radiofrance.fr) into the agent's context through tools like radiofrance_fr__read_page and radiofrance_fr__get_recherche_data_json. This creates a potential surface for indirect prompt injection if the external content contains malicious instructions designed to influence the agent.
Ingestion points: radiofrance_fr__read_page, radiofrance_fr__get_recherche_data_json, and other site-specific tools in SKILL.md.
Boundary markers: None identified in the skill instructions to delimit untrusted web content.
Capability inventory: Tools included in the skill permit searching, reading web pages, and using Unbrowse's automation capabilities (unbrowse.run, unbrowse.browse.open) restricted to the radiofrance.fr scope.
Sanitization: Not explicitly mentioned; the skill claims to return 'readable text' and 'verified results' from the site's first-party API.

radiofrance-fr