The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the data-mystic Python package and several standard data science libraries including pandas, numpy, scikit-learn, prophet, matplotlib, seaborn, networkx, scipy, imbalanced-learn, and ydata-profiling from the official PyPI registry.
[COMMAND_EXECUTION]: The skill defines and executes several CLI commands (e.g., data-mystic correlate, data-mystic anomalies, data-mystic forecast) to perform statistical analysis and pattern recognition on local data files.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its data processing nature.
Ingestion points: The skill ingests external data from files such as CSV, JSON, and system logs (e.g., /var/log/app/events_2024.jsonl) via the --input flag defined in SKILL.md.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore potential commands embedded within the data being analyzed.
Capability inventory: The skill possesses the capability to execute complex data analysis commands and generate reports that are then read by the agent.
Sanitization: While the skill documentation mentions validation of file existence and schema (Step 1 in SKILL.md), it lacks specific content sanitization to filter out malicious instructions hidden within the datasets.

data-mystic