The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads MCP server configurations and tool definitions from the vendor's official domain at unbrowse.ai. This is standard behavior for connecting to a remote MCP service.
[DATA_EXFILTRATION]: User queries are sent to unbrowse.ai to be processed and forwarded to timeanddate.com. This network communication is the primary intended function of the skill and uses the vendor's own infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect injection as it retrieves and processes content from external web pages.
Ingestion points: Data is ingested through the timeanddate_com__get_worldclock and unbrowse.scrape tools which return page titles, text, and links.
Boundary markers: The skill instructions include a directive to "Report only what a succeeded result says," providing a basic guideline for the agent when handling results.
Capability inventory: The skill includes tools for searching, scraping, and performing actions (via unbrowse.browse.act) within the scope of the target site.
Sanitization: There is no specific mention of client-side sanitization, as the vendor service claims to return verified results from first-party API replays.
[CREDENTIALS_UNSAFE]: The documentation provides explicit security guidance for handling API keys, advising users to store the $UNBROWSE_API_KEY in a secret manager and never commit it to configuration files.

timeanddate