Back to skill

Security audit

insync-mcp

Security checks for vulnerabilities and agentic risk

Overview

This patient-portal reader needs review because its raw command permits account-changing requests and unfiltered access to sensitive responses despite its read-only description.

Prefer a version that restricts requests to reviewed read operations and consistently filters sensitive responses. Until then, do not rely on the read-only label: raw requests can invoke account-changing endpoints, and opening unread messages may mark them read. Pair only the intended browser and share only information needed for your question. All four bundled files were inspected; the external CLI, bridge, separate MCP server, and live portal behavior were not verified.

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
Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (5)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

Despite describing the skill as read-only, the documentation exposes arbitrary authenticated endpoint calls through raw GET and POST operations. The prose requirement for confirmation does not establish an enforced restriction, creating a path to unintended patient-account changes; the additional documented profile and document reads are not independently evidence of a vulnerability.

Content

No source excerpt is available for this finding.

The raw command permits arbitrary portal requests, including POST operations, despite the skill's read-only purpose.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The raw branch accepts caller-selected paths and JSON bodies and sends authenticated GET or POST requests without an endpoint allowlist or confirmation check. This defeats the script's read-only claim and bypasses output filtering, potentially enabling account changes or disclosure of sensitive fields that curated commands deliberately omit.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: network.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

The raw command exposes arbitrary portal endpoint calls beyond the manifest's enumerated read operations.

Medium
Category
Excessive Agency
Confidence
96% confidence
Finding

The advertised raw interface bypasses the curated command set and returns endpoint responses without projection. In this healthcare context, that can expose the password hashes and sign-in tokens explicitly identified elsewhere in the document, as well as permit calls beyond the stated read-only purpose.

Content

No source excerpt is available for this finding.

The documents and profile commands retrieve patient information outside the manifest's declared scope.

Medium
Category
Excessive Agency
Confidence
96% confidence
Finding

The manifest lists appointments, medications, allergies, messages, pending forms, balances, and badge counts, but these commands additionally retrieve all patient document records and patient demographics. The profile projection retains every nonempty scalar except Image rather than limiting output to information needed for the advertised functions.

Content

No source excerpt is available for this finding.