The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's error taxonomy recommends resolving missing dependencies by executing bash("pip install xxx"). This practice facilitates the installation of arbitrary third-party software from external registries at runtime, which presents a supply chain security risk.
[COMMAND_EXECUTION]: The skill utilizes the bash tool to perform operations including package installation and the execution of a Python command-line utility (python -c "import ast; ...") to validate the AST syntax of modified code files.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from project artifacts and code files.
Ingestion points: Files including artifacts/metrics.csv, artifacts/equity.csv, artifacts/trades.csv, code/signal_engine.py, and config.json are read into the agent context via read_file.
Boundary markers: Absent. The skill provides no delimiters or instructions to ignore potentially malicious content within the files it processes.
Capability inventory: The skill possesses the capability to execute shell commands via bash and modify project files using edit_file.
Sanitization: Absent. There is no evidence of validation or filtering performed on the content of the read files before it influences the agent's diagnostic logic or subsequent code modifications.

backtest-diagnose