The Agent Skills Directory

[SAFE]: The skill provides utility scripts for data transformation and validation using local processing without network access.
The logic is implemented in standalone Python scripts using standard libraries like pandas and pdfplumber.
No suspicious network operations or credential harvesting patterns were identified.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from external laboratory instrument exports (CSV, PDF, Excel).
Ingestion points: The load_table function in scripts/convert_instrument_data.py reads user-supplied files into memory.
Boundary markers: While the skill does not explicitly provide prompt delimiters for the converted output, the transformation is handled by deterministic Python code rather than LLM-based extraction, which limits the influence of malicious data on the agent's reasoning.
Capability inventory: The skill is restricted to file system operations (reading input and writing standardized output) and lacks network or shell execution capabilities that could be abused via injected data.
Sanitization: The script uses the to_number function to strictly validate and convert measurement data into numeric types (int/float), effectively stripping non-numeric control characters from these fields.
[EXTERNAL_DOWNLOADS]: The documentation references the allotropy library as an optional dependency.
SKILL.md and scripts/convert_instrument_data.py provide instructions for the user to manually install the package via pip if vendor-certified parsing is required.
The skill does not automate this installation or perform runtime code downloads.

tooluniverse-lab-data-standardization