The Agent Skills Directory

[SAFE]: The skill serves as a technical reference and provides utility scripts for handling binary well log data. The implementation utilizes established domain-specific Python libraries and follows secure development practices for local file manipulation.
[INDIRECT_PROMPT_INJECTION]: The skill processes external binary log files, which is a necessary component of its data-loading role. This presents a potential surface for indirect injection if metadata within those files were to contain malicious instructions.
Ingestion points: The dlisio.dlis.load() function is used in SKILL.md code examples and the scripts/dlis_to_las.py conversion script to read external log files.
Boundary markers: The skill does not explicitly use delimiters or instructions for the agent to ignore text embedded in the log data, which is standard for specialized binary parsers.
Capability inventory: Capabilities are strictly limited to file read and write operations for specific formats (DLIS and LAS) and data processing with NumPy/Pandas. No capabilities for network access, subprocess execution, or dynamic code evaluation (eval/exec) were found.
Sanitization: The skill relies on the parsing logic of the dlisio library to process binary data structures, preventing raw data from being directly executed or interpreted as shell commands.

dlisio