The Agent Skills Directory

[Data Exposure & Exfiltration] (LOW): The skill accesses highly sensitive health data, including medical history, medication records, and allergy information (e.g., data/index.json, data/medications.json, data/allergies.json). While no external network tools are explicitly enabled, the ingestion of PHI into the agent context presents a privacy risk if the agent's output is logged or redirected.
[Indirect Prompt Injection] (LOW): The skill processes data from multiple external sources that could be influenced by a user or attacker to override agent behavior.
Ingestion points: Data enters through data-example/fitness-tracker.json, data-example/sleep-tracker.json, and data-example/mental-health-tracker.json.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the skill definition.
Capability inventory: The agent has Read, Write, Grep, and Glob capabilities, which could be exploited if an injection is successful.
Sanitization: There is no evidence of data sanitization or validation for the health tracker files.
[Dynamic Execution] (LOW): The skill documentation mentions calling an external script scripts/generate_ai_report.py for report generation. Although a Python execution tool is not in the allowed-tools list, the reliance on external scripts for core functionality introduces a secondary execution path that is not fully transparent in the skill's defined toolset.

ai-analyzer