The Agent Skills Directory

[COMMAND_EXECUTION]: The documentation includes command-line instructions for executing LabTalk scripts and registering the OriginPro COM server using system-level executables (e.g., Origin64.exe -r and taskkill /F /IM Origin64.exe).
[PRIVILEGE_ESCALATION]: Troubleshooting sections recommend running commands in an 'elevated command prompt' or 'as Administrator' to register COM components (Origin.exe /regserver), which involves acquiring higher system permissions.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data sources including CSV files and OriginPro project archives (.opju).
Ingestion points: CSV files processed in batch_origin_processor.py and project files in inspect_opju.py.
Boundary markers: None identified; data is read directly into dataframes or parsed as XML/Zip archives.
Capability inventory: File system writes (PDF/PNG export), worksheet creation, and analytical fitting via the originpro API.
Sanitization: Standard library parsing (Pandas, zipfile) is used, but there is no explicit sanitization to prevent malicious instructions embedded in data headers or comments from influencing agent behavior.

originpro