Back to skill

Security audit

1ToolRing

Security checks for vulnerabilities and agentic risk

Overview

The verdict is benign with medium confidence: scanners report behavior consistent with the skill’s stated purpose, alongside a data-sharing caution; one referenced local path was missing, so the evidence is incomplete.

Consider whether you are comfortable sending task descriptions and tool inputs to an external broker and providers, and keep confidential details out of those requests. The verdict is benign, but confidence is medium and coverage was partial because a referenced local path was missing.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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 (2)

The workflow sends task descriptions and tool inputs to external services without a privacy warning.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The workflow directs the agent to send task descriptions and tool inputs to an external broker and third-party providers without instructions to minimize sensitive data or obtain consent for sensitive disclosures. External transmission is expected for this skill, so this is not evidence of malicious exfiltration; however, its broad use cases could lead to confidential task details or personal information being unnecessarily included in requests.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% 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 60)May include surrounding context.

md
- Prices are per call, per result or per unit, shown as `{"usd", "idr"}`. The most a run can cost is held when it starts and settled to the real cost. The `X-ToolRing-Cost-USD` and `X-ToolRing-Cost-IDR` headers carry the amount actually charged.
- Failed or provider-error runs cost 0.
- Spend small amounts without asking: run it when one call costs Rp1.000 (about $0.05) or less and the whole request stays under Rp5.000 (about $0.30). Ask first when a single run costs more, the request would pass Rp5.000 in total, or it would use more than half of the remaining balance. When done, state the cost once in a few words, for example "Cost Rp96".
- If the balance is too low (402 INSUFFICIENT_BALANCE), tell the person to top up at https://app.1toolring.com/billing (from Rp10.000 by Indonesian bank virtual account or Alfamart/Indomaret, or from $1 by card). Do not retry until they have.

## 5. Errors