The Agent Skills Directory

[SAFE]: The skill is a standard data processing utility for petrophysical applications. It relies on the well-known and reputable lasio Python library for file operations.
[SAFE]: All scripts (validate_las.py, las_to_csv.py, merge_curves.py) follow standard data science patterns using numpy, pandas, and pathlib for file handling without performing dangerous system calls or network operations.
[INDIRECT_PROMPT_INJECTION]: The skill processes external LAS files which represents a potential attack surface. However, the risk is negligible as LAS files are primarily structured numeric data, and the skill does not provide the agent with high-privilege capabilities (like shell access or network exfiltration) to act on any embedded textual data found in well headers.
Ingestion points: LAS files are read via lasio.read() in all provided scripts.
Boundary markers: None specific to prompt injection; the format itself provides structural boundaries (~WELL, ~CURVE, ~ASCII).
Capability inventory: Local file writing (las.write, df.to_csv) and data manipulation with pandas/numpy.
Sanitization: validate_las.py performs format validation and data quality checks (monotonicity, null counts).

lasio