The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill fetches content from the hometeamns.sg website, which could contain instructions intended to manipulate the agent.
Ingestion points: The tools hometeamns_sg__read_page and hometeamns_sg__get_search (SKILL.md) ingest external web content.
Boundary markers: No delimiters or warnings are used to separate external content from agent instructions.
Capability inventory: The skill possesses the ability to execute high-level tasks via unbrowse.run and perform automated browser actions, including autofill from a password manager, via unbrowse.browse.act (SKILL.md).
Sanitization: No evidence of sanitization or validation of the fetched web content was found.
[EXTERNAL_DOWNLOADS]: The skill connects to a remote Model Context Protocol (MCP) server provided by the vendor.
Evidence: The server.json file and SKILL.md reference a remote connection to https://unbrowse.ai/mcp/hometeamns.sg.

hometeamns-sg