The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements an optional conversational interface (llm_trading_assistant.py) that summarizes market analysis results for the user. This architecture presents an indirect prompt injection surface where unvalidated data from external cryptocurrency exchange APIs could influence the behavior of the assistant.
Ingestion points: llm_trading_assistant.py processes user messages and technical analysis summaries derived from external exchange ticker data.
Boundary markers: Absent; analysis summaries are directly interpolated into the system context without explicit delimiters or instructional guarding.
Capability inventory: The assistant is capable of generating conversational explanations and trading recommendations using OpenAI or Anthropic APIs.
Sanitization: No filtering or escaping is applied to ticker data or symbol summaries before they are processed by the LLM.
[EXTERNAL_DOWNLOADS]: The skill connects to well-known cryptocurrency exchanges (Binance, Coinbase, Kraken) using the CCXT library to fetch real-time market data for analysis.
Source: Established cryptocurrency exchange APIs.
Usage: Read-only OHLCV (Open, High, Low, Close, Volume) data used for technical indicators and mathematical simulations.
Risk level: Safe, as data is fetched from well-known services and processed within a 6-stage validation pipeline.

cryptocurrency-trader