The Agent Skills Directory

Indirect Prompt Injection Surface: The skill processes user-provided inputs (subject, example_data) to determine the schema and patterns for the generated notebook code. (1) Ingestion points: Data enters the agent context through the subject and example_data parameters in SKILL.md. (2) Boundary markers: The instructions do not specify explicit delimiters for interpolating user inputs into the code-generation logic. (3) Capability inventory: The skill possesses the ability to create directories, create and edit Jupyter notebooks, and execute Python code cells using specialized tools. (4) Sanitization: The skill implements a mandatory security gate that requires a validated preflight record from a separate dataops skill before any operations begin.
Dynamic Execution: The skill generates Python code cells in a Jupyter notebook and executes them to perform data generation. This behavior follows established templates and is gated by pre-execution validation steps.
Environment Verification: The skill performs a version check for the Tesseract OCR engine (tesseract --version) using terminal tools when image processing is requested. This is a routine check for software availability.
Standard Library Installation: The generated notebook environment installs several well-known and standard Python libraries (pandas, numpy, matplotlib, seaborn, scipy) to support its core data generation and analysis functions.

synth-data-generate