The Agent Skills Directory

[SAFE]: The skill facilitates delivery management tasks using local Python scripts and markdown instructions. A technical audit of the provided scripts (delivery_metrics_tracker.py, dependency_mapper.py, and risk_register.py) confirms they are limited to data processing and reporting tasks using standard Python libraries. No evidence of malicious network communication, credential theft, or unauthorized file modifications was found.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests and processes user-supplied JSON data that is then presented to the agent for analysis.
Ingestion points: Data files loaded via command-line arguments in scripts/delivery_metrics_tracker.py, scripts/dependency_mapper.py, and scripts/risk_register.py.
Boundary markers: Absent. There are no explicit delimiters defined in the skill instructions to separate external tool outputs from system prompts.
Capability inventory: The included scripts do not contain any dangerous capabilities such as eval(), exec(), subprocess calls, or network operations.
Sanitization: The scripts parse JSON structure but do not sanitize or filter the content of natural language fields before they enter the agent's context.

delivery-manager