ClawHub
Back to skill

Security audit

Coding Agent

Security checks for vulnerabilities and agentic risk

Overview

ClawScan returns a benign verdict with medium confidence: the reported concern is that worker launches may send task prompts and project content to external AI providers without a privacy warning, and the scan was only partial.

Before installing, consider whether you are comfortable with task prompts and project content being sent to the configured external AI provider. The scan did not confirm that such data is approved for that provider, and the analysis was partial because two local references could not be resolved.

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
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Worker launches lack a privacy warning about sending task prompts and project context to external AI providers.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

The skill directs workers to use external AI services without requiring disclosure of the data transfer or verification that the task and repository are approved for the configured provider. Prompts and project content read by those workers may therefore expose confidential code, personal data, or secrets without informed authorization. External AI delegation is the skill's stated purpose, which reduces evidence of deceptive intent, but its worktree and authentication safeguards do not address provider-side data disclosure.

Content

No source excerpt is available for this finding.