The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses highly sensitive personal information, including medical history records (data/index.json), medication lists (data/medications.json), allergy data (data/allergies.json), and mental health tracking results such as PHQ-9 and GAD-7 scores (data-example/mental-health-tracker.json). The processing of this sensitive health data constitutes a potential exposure risk.- [COMMAND_EXECUTION]: The skill instructions direct the agent to execute an external Python script (scripts/generate_ai_report.py) to generate interactive HTML reports. Although the allowed-tools field in the skill metadata only specifies Read, Grep, Glob, and Write, the procedural steps explicitly call for script execution, indicating an attempt to perform operations that may exceed defined tool constraints.- [PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection by processing external data from multiple sources without validation.
Ingestion points: Data is read from several JSON files in the data/ and data-example/ directories, including fitness, sleep, and nutrition trackers.
Boundary markers: No boundary markers or clear instructions are provided to the agent to distinguish between data content and system instructions.
Capability inventory: The skill possesses file-reading (Read), file-writing (Write), and script execution capabilities.
Sanitization: There is no evidence of sanitization, escaping, or filtering of the content retrieved from external data files before it is processed or used to generate reports.

ai-analyzer