Back to skill

Security audit

ElevenReader

Security checks for vulnerabilities and agentic risk

Overview

Review is recommended because setup can execute unverified remote installers, and the safety rules incorrectly classify an audio-upload action as read-only.

Install the CLI separately using a reviewed, integrity-verified method rather than letting a failed reading request trigger installation. Before using read_text, explicitly authorize the text and its upload, especially for private material, and check the provider's storage controls. The review concerns follow directly from the skill's instructions; no malicious payload was established, and the external installer contents remain unverified.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (5)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The supplied command executes an external installer rather than searching or reading ElevenReader data. No connection between this installer and the declared ElevenReader functionality is established in the supplied code. The downloaded script's contents are not provided, so its further actions cannot be determined.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
98% confidence
Finding

Piping an HTTPS response directly into bash executes mutable remote code without checking a pinned version, signature, or independently trusted digest. If the installer distribution is compromised, the script could access files and credentials or modify the environment with the invoking process's privileges. The OOMOL domain is relevant to the documented dependency, but relevance and HTTPS do not establish script integrity.

Content

Scanner excerpt · SKILL.md (reported line 59)May include surrounding context.

The untagged read_text action uploads generated audio, contradicting the claim that all untagged actions are read-only.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The untagged read_text action generates speech and uploads the resulting audio to connector transit storage, despite the blanket claim that untagged actions are read-only and safe to run directly. This can bypass the skill's confirmation rule for state-changing operations and create an externally stored spoken copy of potentially sensitive text. Speech generation fits the skill's purpose, but that does not make the upload read-only or establish informed authorization.

Content

No source excerpt is available for this finding.

The fallback setup instructs execution of downloaded installer scripts without warning about running remote code or modifying the system.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

The missing-CLI fallback directs the agent to execute downloaded installer code without a separate approval gate or warning about local system changes. A routine dependency error can therefore expand a reading task into local software installation with the agent's execution privileges. This overlaps with SC2 rather than representing an independent exploit path.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 80)May include surrounding context.

md
- **`OOMOL_INSUFFICIENT_CREDIT`** — OOMOL billing stop. Recharge at `https://console.oomol.com/billing/token-recharge` before retrying.

- **HTTP 402 without `OOMOL_INSUFFICIENT_CREDIT`** — the third-party account itself has no active plan or credit. Report the provider error message as is; do not ask the user to recharge OOMOL.

## Resources