The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill fetches market details, such as market questions and selection names, from the Betfair API and includes them in the output shown to the agent without any sanitization or isolation markers. This creates a surface where malicious instructions embedded in market names or descriptions could potentially influence the agent's behavior.
Ingestion points: Data is ingested from the external Betfair API in the handleMarkets and handleMarket functions within index.ts.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the fetched market data as untrusted content.
Capability inventory: The skill possesses sensitive capabilities including placing back/lay orders (financial transactions) and retrieving account balance and position data across several functions in index.ts.
Sanitization: The code directly interpolates strings like market.question and o.name into the markdown response without escaping, filtering, or validation.

betfair