Back to skill

Security audit

CNB

Security checks for vulnerabilities and agentic risk

Overview

The CNB features are clearly scoped, but setup directs the agent to execute downloaded software without a clear approval step.

Review and install the CLI yourself before enabling this skill, and require approval for any fallback software installation. Check the CLI's bundled-skill installation behavior and grant only the CNB permissions you need. CNB changes already require confirmation. No malicious intent was established; the concern is insufficient control over setup, whose downloaded code could not be inspected.

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 (4)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The declared purpose concerns CNB data operations, whereas the supplied command executes an installer from a different service. No visible code establishes that this installation supports CNB access. The remote script's contents are not supplied, so its further behavior cannot be determined.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
97% confidence
Finding

The command immediately executes a mutable remote script without version pinning or independent integrity verification; the PowerShell alternative has the same property. Although the vendor domain and prerequisite-installation context appear legitimate, compromise of the distribution endpoint could execute arbitrary code with the invoking user's access to local files and credentials.

Content

Scanner excerpt · SKILL.md (reported line 68)May include surrounding context.

The setup workflow executes downloaded installation scripts without warning that this runs remote code and modifies the local system.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

The fallback directs installation when the CLI is missing without explaining that downloaded code will execute locally or requiring approval for local system changes. Its conditional nature limits exposure, but the user-confirmation rules cover CNB writes rather than installation, leaving a consent gap if the agent can execute these commands.

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 89)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