The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests external data from market signals and exchange API responses to make trading decisions.
Ingestion points: Signal data from a market-monitor skill and JSON responses from external exchange API endpoints (Binance and OKX).
Boundary markers: The skill implements a mandatory human-in-the-loop confirmation step before any execution and includes a specific risk check block.
Capability inventory: The skill has the capability to perform network POST requests to financial exchanges to execute trades and manage assets.
Sanitization: Risk checks are performed against a whitelist of trading pairs (allowedPairs) and transaction limits defined in a configuration file.
[EXTERNAL_DOWNLOADS]: The skill performs network operations targeting well-known financial services.
Fetches order status and executes trades via Binance's official API (api.binance.com and fapi.binance.com).
Communicates with OKX's official API (api.okx.com) for trade execution and algorithmic order management.
[DYNAMIC_EXECUTION]: The skill utilizes a dynamic loading mechanism to select the appropriate implementation file (impl/{exchange}.md) based on the user-provided exchange parameter. Loading implementation logic from computed paths at runtime can lead to path traversal vulnerabilities if the input parameter is not strictly validated against a predefined list of supported exchanges.

trade-executor