The Agent Skills Directory

[SAFE]: The skill follows secure data handling practices by implementing mandatory SHA256 integrity verification for all input data files against a provided provenance record.
[SAFE]: File system operations are constrained to user-specified local paths provided via command-line arguments, with explicit checks to prevent overwriting existing data.
[SAFE]: The Python script uses standard, well-vetted scientific libraries (xarray, pandas, numpy, scipy, verde, pyproj) for all calculations and data transformations.
[SAFE]: No network connectivity or remote data fetching is attempted; the skill operates entirely on local datasets as documented.
[INDIRECT_PROMPT_INJECTION]: While the skill ingests external data (CSV/JSON), it implements a strict validation chain:
Ingestion points: Local CSV source and JSON provenance files (scripts/station_pipeline.py).
Boundary markers: Explicit schema validation, column type enforcement, and SHA256 checksum verification.
Capability inventory: Local file writing for NetCDF and JSON reports; no shell execution or network tools.
Sanitization: Rigid parsing of temperature values and coordinate validation ensure the data is not treated as executable instructions.

climate-analysis