The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process large tabular datasets from external sources, which constitutes a vulnerability surface for indirect prompt injection.
Ingestion points: The skill loads data from local files (CSV, HDF5, Parquet) and remote cloud storage (S3, GCS, Azure) using functions like vaex.open() and vaex.from_csv(), as described in references/core_dataframes.md and references/io_operations.md.
Boundary markers: No specific delimiters or instructions for the agent to treat data content as untrusted are documented.
Capability inventory: The skill possesses file writing capabilities (e.g., export_hdf5), cloud network access, and mentions running the vaex-server utility.
Sanitization: The provided references do not include methods for sanitizing external data to prevent it from containing malicious instructions.
[DYNAMIC_EXECUTION]: The skill supports the use of Numba JIT compilation (@numba.jit) for custom performance optimizations and integrates with machine learning frameworks like TensorFlow and Keras, which involve runtime model execution, as documented in references/performance.md and references/machine_learning.md.

vaex