The Agent Skills Directory

[SAFE]: The skill provides a set of quantitative research tools for evaluating market signals. Analysis of the scripts and instructions reveals no malicious intent or security vulnerabilities.
[EXTERNAL_DOWNLOADS]: The skill includes a requirements.txt file referencing standard, reputable Python libraries (numpy, pandas, scipy). These are essential for the skill's mathematical and data processing functions.
[COMMAND_EXECUTION]: The skill includes a Python script scripts/compute_metrics.py which performs purely numerical calculations using the aforementioned libraries. It does not perform any dangerous shell operations, system modifications, or arbitrary code execution.
[PROMPT_INJECTION]: The SKILL.md file contains instructions for the AI agent to maintain the skill's own documentation ("Self-Evolving Skill"). This is a meta-instruction for self-improvement and does not target safety filters or bypass core agent constraints.
[DATA_EXFILTRATION]: No network-enabled code was found in the provided scripts. The skill operates locally on user-provided data files (predictions and actuals in .npy format). References to external GitHub repositories point to the author's own project-level experiment catalogs and are used for documentation purposes.

opendeviation-eval-metrics