The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses highly sensitive personal and medical data files that contain private health information (PHI).
Evidence: The skill is designed to read data/medications.json, data/allergies.json, and data-example/mental-health-tracker.json as part of its core analysis workflow (SKILL.md).
[INDIRECT_PROMPT_INJECTION]: The skill processes health and lifestyle data from external sources that could be used to inject malicious instructions into the AI's analysis or generated reports.
Ingestion points: Data is ingested from external-facing tracker files including fitness-tracker.json, sleep-tracker.json, nutrition-tracker.json, and mental-health-tracker.json (SKILL.md).
Boundary markers: The instructions do not specify any delimiters or safety markers to differentiate between health data and instructions, nor do they include commands to ignore embedded instructions within these data sources.
Capability inventory: The skill possesses the capability to read all local data files and write output to HTML reports via a helper script (SKILL.md).
Sanitization: There is no evidence of input validation, escaping, or sanitization of the data ingested from tracker files before it is processed by the AI or included in interactive reports.
[DYNAMIC_EXECUTION]: The skill relies on the execution of a local Python script to generate its primary output, which introduces a dependency on external code execution logic.
Evidence: Step 8 of the execution instructions calls for the invocation of scripts/generate_ai_report.py to produce HTML reports (SKILL.md).
[METADATA_POISONING]: There is a functional inconsistency between the skill's defined configuration and its operational instructions.
Evidence: The skill's YAML frontmatter restricts allowed-tools to Read, Grep, Glob, and Write. However, the execution steps require running a Python script (generate_ai_report.py), which would typically necessitate a command execution tool (such as Bash or Python) that has been explicitly excluded from the allowed toolset.

ai-analyzer