The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests untrusted metadata, such as token names and social media links, retrieved from blockchain feeds through the XXYY API.
Ingestion points: Metadata is processed in the feed_scan, token_query, and trending_list tools, which retrieve data from the xxyy.io API (referencing SKILL.md and source files in mcp/src/tools/).
Boundary markers: The skill does not explicitly use delimiters or sanitization for these external strings before they are incorporated into the agent's context.
Capability inventory: The skill can execute transactions (buy/sell) and launch new tokens.
Sanitization: No filtering or escaping is performed on external content such as token descriptions or social links.
Mitigation: The risk is well-mitigated by the skill's mandatory requirement for manual user confirmation (via AskUserQuestion) for all write operations, ensuring the user reviews trade details before execution.
[EXTERNAL_DOWNLOADS]: The skill interacts with the XXYY API hosted at https://www.xxyy.io using curl for all its core trading and data retrieval functions.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute curl commands for network communication with the backend trading platform.

xxyy-trade