ClawHub
Back to skill

Security audit

基于AI的SBTI技能测试

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a local SEO planning tool, but it automatically sources a workspace secrets file even though the planner does not appear to need credentials.

Install only if you are comfortable reviewing and controlling the runner script. Create the env file yourself, keep it limited to simple path and scoring settings, do not place unrelated API keys or account credentials in it, and review any backlink outreach manually before acting.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding
The skill instructs the agent to read and write local files and invoke scripts that likely perform network-enabled SEO operations, but it does not declare any permissions or capability boundaries. This creates a mismatch between documented behavior and the platform's trust model, increasing the risk of over-privileged execution, unintended data access, or silent outbound activity if the skill is run in an environment that relies on explicit permission declarations.

Context-Inappropriate Capability

Medium
Confidence
89% confidence
Finding
The script automatically sources an external environment file from a workspace secrets path even though its stated purpose is local SEO backlink planning. Sourcing a shell env file executes shell syntax in the current process, so if that file is modified or attacker-controlled it can run arbitrary commands and expose unrelated secrets to the skill's execution context.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The default prompt initiates a broad SEO workflow with no explicit scope limits, approval checks, or prohibited actions, which can cause an agent to take or recommend manipulative SEO steps beyond intended boundaries. In an SEO context, vague autonomy is riskier because backlink acquisition, anchor manipulation, and relay planning can drift into spammy or policy-violating behavior if not tightly constrained.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The script silently reads a secrets env file from `/root/.openclaw/workspace/.secrets/...` without any indication that an SEO tool will access sensitive credentials. In this skill context, that is more suspicious because backlink planning should not normally require broad secret access, increasing the risk of unnecessary credential exposure and abuse if the environment file contains unrelated tokens.

Static analysis

No suspicious patterns detected.