The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements an LLM-based decision gate in runtime.yaml that processes market data signals via template interpolation ({{signal_piranha_signals}}). This configuration creates a surface for indirect prompt injection.
Ingestion points: Market data (open interest, candles, and order book) is fetched from the platform's MCP tools in scripts/piranha-producer.py and passed to the LLM gate.
Boundary markers: The LLM prompt uses standard placeholder interpolation without explicit escape delimiters or safety instructions regarding the content of the data fields.
Capability inventory: The agent has the capability to execute the OPEN_POSITION action based on the LLM's decision.
Sanitization: The producer script validates the market data and casts relevant signals to specific numeric types (floats) before including them in the emitted signal block.
[EXTERNAL_DOWNLOADS]: The README.md provides installation instructions involving the download of scripts and configuration files from the author's public GitHub repository (github.com/Senpi-ai). These downloads are part of the standard deployment process for the strategy.
[COMMAND_EXECUTION]: The strategy utilizes a long-lived Python daemon script (piranha-producer.py) to monitor assets and generate signals. This script interacts with the local system environment and the trading runtime SDK to perform its duties.

piranha-strategy