The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes scripts that ingest untrusted data from local files (YAML and CSV) for financial modeling.
Ingestion points: The scripts scripts/channel_margin_calculator.py, scripts/channel_mix_optimizer.py, and scripts/partner_tier_economics.py read input from file paths provided via command-line arguments.
Boundary markers: Not present, as the scripts perform mathematical calculations rather than natural language processing.
Capability inventory: The scripts can print results to standard output or write to a user-specified file path using the --output flag. They do not perform network operations or execute subprocesses.
Sanitization: The scripts use a custom, limited YAML parser and standard CSV parsing with basic type conversion (float/int), which effectively mitigates common data-driven execution risks.
[REMOTE_CODE_EXECUTION]: The provided Python scripts are implemented using only the Python standard library (stdlib) and do not download or execute code from external sources.
[DATA_EXFILTRATION]: No network operations (e.g., curl, wget, requests) were found in the scripts or instructions. The skill operates entirely on local files and user input.

channel-economics