The Agent Skills Directory

[SAFE]: No security threats were identified across all 10 threat categories. The skill's instructions and resources are limited to local data science tasks and mathematical modeling.\n- [DATA_EXFILTRATION]: Analysis of the provided Python code and instructions confirms no network communication, external API calls, or access to sensitive system files. Operations are conducted entirely on in-memory DataFrames.\n- [PROMPT_INJECTION]: No attempts to bypass agent safety filters, override system instructions, or use adversarial role-play patterns were detected in the skill's metadata or implementation logic.\n- [PROMPT_INJECTION]: Regarding indirect prompt injection: 1. Ingestion point: historical time-series data (CSV/DataFrames). 2. Boundary markers: missing (standard for data processing). 3. Capability inventory: restricted to Pandas/NumPy mathematical operations; no subprocess, network, or file-write capabilities are present in the provided templates. 4. Sanitization: the framework uses strong typing (Decimal, Enum, Pandas types) which mitigates instruction injection into numerical processing flows. The risk is assessed as safe due to the complete lack of dangerous execution capabilities.

backtesting-frameworks