The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script (scripts/assumption_mapper.py) using python3. The execution is limited to the local logic provided with the skill for processing assumption scores and does not involve unsanitized shell command construction.
[DATA_EXPOSURE]: The skill reads input from CSV files provided at runtime. It does not access sensitive system directories, environment variables, or hardcoded credentials.
[REMOTE_CODE_EXECUTION]: The skill does not perform any network operations or download external scripts. All code execution is confined to the local script file.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external files (CSV) which constitutes an attack surface, but the risk is minimized by the lack of dangerous capabilities.
Ingestion points: The skill reads from assumptions.csv via scripts/assumption_mapper.py.
Boundary markers: No explicit delimiters or instructions to ignore embedded content are provided to the agent for the CSV content.
Capability inventory: The skill is limited to file reading and local script execution; it lacks network access, file writing, or dynamic code evaluation (eval/exec) capabilities.
Sanitization: The script performs basic type validation by ensuring risk and certainty values are floats between 0 and 1.

product-discovery