The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the zig build run pattern. These commands are used to interact with a local trading engine for validating strategies, running backtests, and performing parameter sweeps.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes external TOML strategy files and JSON payloads. If a user provides a malicious strategy file, it could theoretically attempt to influence the agent's interpretation of results or exploit the underlying engine's expression parser (logic.formula).
Ingestion points: Reads strategy configuration from ../strategies/*.toml and receives JSON strings in API mode.
Boundary markers: None explicitly defined in the instructions for the agent to use when parsing the strategy files.
Capability inventory: Performs file system reads/writes (via the engine's export feature) and subprocess execution via zig.
Sanitization: The skill does not explicitly instruct the agent to sanitize or validate the content of the TOML files before passing them to the build command.

quaq-backtest