The Agent Skills Directory

[Data Exposure] (LOW): The skill reads highly sensitive health-related files including medications.json, allergies.json, and index.json. This behavior is expected given the skill's primary function of health risk prediction and no network exfiltration patterns were identified.
[Indirect Prompt Injection] (LOW): There is an attack surface for indirect prompt injection as the skill processes external health data files.
Ingestion points: Reads data from data/*.json and data-example/*.json (e.g., fitness, sleep, and nutrition trackers).
Boundary markers: No delimiters or warnings to ignore embedded instructions are present in the processing logic.
Capability inventory: The skill uses Read, Grep, Glob, and Write tools.
Sanitization: No explicit sanitization or validation of the input data is described before processing.
[Command Execution] (SAFE): While Step 8 of the execution instructions mentions running a Python script (scripts/generate_ai_report.py), the allowed-tools header correctly limits the agent to file operations (Read, Grep, Glob, Write), effectively preventing the execution of arbitrary scripts or commands unless a Python execution tool is separately provided by the environment.

ai-analyzer