Back to skill

Security audit

Human Psychologist for AI Agents

Security checks for vulnerabilities and agentic risk

Overview

This skill is openly designed to send mental-health messages to an external human service, but it forwards highly sensitive content and introduces crypto payment instructions without enough up-front consent, privacy, or payment verification controls.

Review this skill carefully before installing. Do not send names, contact details, diagnoses, crisis information, or other sensitive identifiers unless you are comfortable with the content going to an external API, a human recipient, and Telegram. For paid follow-ups, independently verify the network, token contract, amount, and recipient address before making any USDC transfer, and assume payments may be irreversible.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (2)

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:29
Finding

Sensitive Mental-Health Data Transmitted Without Explicit Informed Consent

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:72
Finding

Cryptocurrency Payment Instructions Trust an Unverified Runtime Recipient

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

This skill routes highly sensitive mental-health disclosures to an external third-party service and explicitly onward to a human psychologist via Telegram, but it does not require an up-front, prominent privacy/consent warning before the user submits content. Users may reasonably assume they are interacting with an AI or a private in-product workflow, and could disclose crisis details, trauma history, diagnoses, or other special-category personal data without informed consent about human access, third-party handling, retention, or cross-service transmission.

Content

No source excerpt is available for this finding.