The Agent Skills Directory

[PROMPT_INJECTION]: The skill defines a prompt template in SKILL.md that ingests untrusted user data via variables such as {{expr}} and {{task}}. Evidence Chain: 1. Ingestion points: SKILL.md (task, attachments, expr). 2. Boundary markers: Uses simple double-bracket and block delimiters (e.g., {{#task}}). 3. Capability inventory: Python execution for data manipulation and chart generation via Matplotlib. 4. Sanitization: Absent. This represents a surface for indirect prompt injection where a user could embed instructions within a factor expression.
[SAFE]: The core backtesting algorithm in references/algorithm.md and diagnostic code in references/diagnostic-charts.md use standard libraries like Pandas and Matplotlib for financial modeling. No unauthorized network activity, credential harvesting, or persistent mechanisms were identified.
[SAFE]: All external references, such as community images, point to the official repository of the author (quantskills) on a well-known platform (GitHub).

backtest