The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions explicitly direct the agent to fetch untrusted external data and incorporate it into the game environment, creating an attack surface for indirect prompt injection.
Ingestion points: External data enters the agent context via the socialdata MCP tools (get-profile, search-tweets) and web search results, as described in the 'Data-Inspired Design' section of SKILL.md.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when processing or quoting text from tweets and articles.
Capability inventory: The skill utilizes several tools to modify the game state, including create_zone, update_zone_npcs, update_zone_signs, and delete_zone.
Sanitization: The skill lacks instructions for sanitizing, escaping, or validating the external content before it is interpolated into NPC dialogue, sign text, or building labels.

world-builder