ClawHub
Back to skill

Security audit

clawguard

Security checks for vulnerabilities and agentic risk

Overview

The skill coherently provides remote security scans, with disclosed data sharing and paid reports and no hidden local data collection.

Submit public repository paths only. Decide whether to allow paid reports: the instructions request them automatically after REVIEW or BLOCK and suggest using a configured wallet. Keep payment approval enabled if you want control over charges, and treat a PASS as advisory.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Code capability 'network' detected in scripts/clawguard.py but not covered by declared permissions.

High
Category
MCP Least Privilege
Confidence
88% confidence
Finding

The client performs outbound HTTPS requests through urllib.request.urlopen, which confirms the network capability reported as undeclared. If permissions are intended to govern user consent or execution, this omission allows an external service to receive the supplied repository identifier without a corresponding declared authorization. Networking is central to this skill's stated purpose, and the code does not explicitly upload local files or execute payments, making this a permission-transparency issue rather than evidence of malicious exfiltration.

Content

No source excerpt is available for this finding.

Code capability 'shell' detected in scripts/clawguard.py but not covered by declared permissions.

High
Category
MCP Least Privilege
Confidence
75% confidence
Finding

The skill uses 'shell' capability that is not covered by the declared 'allowed-tools' preapproval. Review the host runtime's actual enforcement policy; runtime permissions are unknown unless separately verified, and this declaration alone does not establish a permission ceiling.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.