The Agent Skills Directory

Local Script Execution: The skill is configured to execute a local pre-flight script (python3 skills/user-context/scripts/user_context_preflight.py) to initialize context. This is a standard architectural pattern for modular agent skills to ensure necessary state is loaded before proceeding with analysis.
Structured Data Processing: The analytical engine (implemented in scripts/run_plan.py) processes user-provided CSV and XLSX files according to strict schemas defined in the references/ directory. It employs robust data handling via the pandas library, including numeric coercion and safe type casting, which reduces the risk of data-driven errors.
File Integrity and Auditing: The skill includes built-in security and quality assurance features, such as SHA256 hashing of input files and the generation of a run manifest and QA reports. These mechanisms ensure that the data used for financial modeling is traceable and consistent.
Secure Configuration Management: The skill utilizes PyYAML with safe loading practices for sector-specific KPI configurations. It avoids dangerous dynamic execution patterns (like eval or exec) and performs all operations within localized directories, maintaining a clean execution environment.
Standard Dependencies: The Python environment relies on widely-vetted, well-known libraries such as pandas, numpy, and PyYAML, ensuring a stable and secure runtime profile.

earnings-preview