Security audit

Graph Interpretation

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a graph-explanation helper, but it enters clinical decision-support and patient-facing medical interpretation without clear medical-use limits or required expert review.

Review this carefully before installing for healthcare work. It may be acceptable for drafting figure captions or research summaries, but users should not rely on it for diagnosis, treatment choices, patient counseling, or clinical decisions without qualified professional review and source-data verification.

A.I.G

Vulnerability Patterns

Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Missing User Warnings

High

Confidence: 99% confidence
Finding: Advertising generation of 'Clinical decision support' is materially riskier than general graph interpretation because it implies downstream use in patient-care decisions. Without strong restrictions, human-review requirements, and a prohibition on autonomous clinical recommendations, users could rely on model-generated output in ways that harm patients through misinterpretation of evidence or inappropriate treatment choices.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The skill is explicitly positioned for clinical research and clinical reporting, but it does not warn users that generated interpretations may be inaccurate, incomplete, or unsuitable as medical advice or sole decision support. In a healthcare context, omission of such guardrails can cause users to over-trust outputs and apply them to patient care or scientific communication without appropriate expert verification.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The clinician- and patient-facing examples state treatment benefits and suggest care actions in plain language without any safety disclaimer or uncertainty framing beyond the example statistics. This can normalize the model producing persuasive medical summaries that users may mistake for validated clinical guidance, especially when directed at patients or clinicians.

Static analysis

No suspicious patterns detected.