Back to skill

Security audit

agent-data-tools

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because its MCP setup puts a prepaid API key in a URL, which could expose the key through logs or diagnostics.

Consider the key-in-URL concern before installing. The scan recommends using an authorization header instead and advises minimizing task details sent to the external service, especially confidential or personal information. The scan was partial because it could not resolve a local path-like reference, so its evidence is incomplete.

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
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

The tool-discovery workflow sends task details to a third-party service without a privacy warning.

Medium
Category
Output Handling
Confidence
86% confidence
Finding

The workflow instructs the agent to tell the external finder which job it needs performed, potentially transmitting user-provided task context. Although the file identifies the destination and HTTP mechanism, it does not warn against including confidential or personal information in discovery requests or subsequent tool inputs; its secrecy guidance covers API keys only.

Content

No source excerpt is available for this finding.

The MCP connection instructions put the prepaid API key in a URL despite explicit assurances that keys stay out of URLs.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

The MCP fallback embeds a prepaid bearer key in the URL path, where it can be captured by access logs, proxy telemetry, client configuration exports, or diagnostic reports. Anyone obtaining that URL could use the key and consume its prepaid credits; the instruction to treat the URL as secret reduces accidental sharing but does not prevent infrastructure logging.

Content

No source excerpt is available for this finding.