The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Numba library to perform Just-In-Time (JIT) compilation of user-defined functions and applies them dynamically to datasets. Evidence: references/performance.md demonstrates @numba.jit and df.apply for runtime-compiled calculations.
[COMMAND_EXECUTION]: The skill provides instructions for executing the vaex-server command to host datasets over a network, which involves shell-level execution. Evidence: references/io_operations.md documented CLI usage of vaex-server.
[EXTERNAL_DOWNLOADS]: The skill enables downloading and processing data from well-known cloud providers such as Amazon S3, Google Cloud Storage, and Azure Blob Storage. Evidence: references/io_operations.md includes configuration for s3://, gs://, and az:// data access.
[PROMPT_INJECTION]: The skill provides a significant surface for indirect prompt injection as it processes large volumes of data from external files and cloud sources without boundary markers or sanitization.
Ingestion points: Data is loaded from CSV, Parquet, and HDF5 files via vaex.open and vaex.from_csv in references/io_operations.md.
Boundary markers: No delimiters or protective instructions are provided to the agent to distinguish data content from instructions.
Capability inventory: The skill maintains access to file-writing, network-based data retrieval, and dynamic code execution (JIT).
Sanitization: No validation or sanitization mechanisms for imported data are specified in the documentation.

vaex