Security audit
Security checks for vulnerabilities and agentic risk
This allergy-tracking skill operates locally and mostly as advertised, but users should treat its medication timing output as informational and confirm choices with a clinician or pharmacist.
Install only if you are comfortable keeping allergy symptom data in local JSON files and using the tool as a planning aid. Do not rely on it for diagnosis or urgent symptoms, and confirm medication choices and timing with a pharmacist or clinician, especially for children, pregnancy, other medical conditions, or any breathing symptoms.
The skill instructs users to read and write local JSON log files via command-line examples, but the manifest does not declare any tool scope such as allowed-tools or permissions. This creates an authorization/expectation gap: a host agent may permit filesystem access implicitly or more broadly than intended, increasing the chance of unintended local file access beyond the allergy log workflow.
Suspicious Unicode normalization or mixed-script content
Data from a source is assigned to a variable that is later passed to a sink, creating a variable-mediated taint flow.
entries.append({"date": args.date, "score": args.score,
"systems": systems, "note": args.note or ""})
entries.sort(key=lambda e: e["date"])
p.write_text(json.dumps(entries, indent=2) + "\n")
print("Logged %s score=%d systems=%s" % (args.date, args.score,
"+".join(systems) or "-"))
return 0
The script gives concrete medication-timing advice such as starting antihistamines immediately and suggesting pre-medication schedules, while only providing a weak disclaimer. In a health-related skill, this can cause users to act on personalized-seeming medical guidance without clinician review, which is especially risky for users with contraindications, comorbidities, pregnancy, pediatric use, or severe respiratory symptoms.