The Agent Skills Directory

[SAFE]: The skill is focused on providing professional guidance and local data processing for quality management systems. All scripts and documentation align with the stated purpose of assisting in CAPA investigations.
[COMMAND_EXECUTION]: The skill references two local Python scripts (scripts/capa_tracker.py and scripts/root_cause_analyzer.py) for managing CAPA data and performing root cause analysis. These scripts use only Python standard library modules and perform safe data processing without any dangerous system calls or shell execution of untrusted data.
[DATA_EXFILTRATION]: Analysis of the Python scripts and markdown files shows no network operations (such as HTTP requests), no access to sensitive system paths (like credentials or SSH keys), and no hardcoded secrets. All data processing is performed locally on user-provided JSON files.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external JSON data describing CAPA items and investigations.
Ingestion points: Data enters the environment via the --capas and --data arguments in the provided Python scripts.
Boundary markers: None are explicitly used when the agent reads the output of these tools.
Capability inventory: The scripts are restricted to data calculation and text report generation; they lack file-write or network capabilities.
Sanitization: The scripts parse JSON but do not sanitize the text content for potential prompt injection. However, since the output is primarily for human/agent review of quality metrics, the risk is negligible.

capa-officer