Back to skill

Security audit

AgentEarth

Security checks for vulnerabilities and agentic risk

Overview

AgentEarth operations have clear approval checks, but setup executes unverified installers and leaves broader changes to agent skills undisclosed.

Review the CLI setup before enabling this skill. Prefer a separately reviewed, verified CLI installation and disable startup skill synchronization unless you want it changing other agents' skills. AgentEarth execution can delete external resources and consume credits, so inspect the requested target and effect before approving. The installer contents and hosted implementation were not verified; no malicious behavior was established.

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
97% confidence
Finding

The supplied chunk performs remote script execution rather than AgentEarth data operations. Nothing supplied establishes that the Oomol installer is a supporting dependency for AgentEarth. The downloaded script's contents are unavailable, so its further behavior cannot be determined.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
98% confidence
Finding

Piping the HTTPS installer directly into bash gives the remote response immediate code execution with the user's privileges. HTTPS protects transport but does not protect against compromise of the installer publisher or hosting infrastructure; this is the same underlying installation weakness identified by SQP-2.

Content

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

First-time setup instructs immediate execution of downloaded installers without a remote-code execution warning.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

Both setup examples execute remotely fetched code immediately, without inspection, version pinning, or independent integrity verification. Although installation is a legitimate prerequisite and only triggered when the CLI is missing, a compromised distribution endpoint could execute arbitrary code with the invoking user's privileges; the underlying issue is unverified execution, not merely the missing warning.

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